Professional Documents
Culture Documents
ibm.com/redbooks
Draft Document for Review November 14, 2006 3:49 pm 6786edno.fm
October 2006
SG24-6786-02
6786edno.fm Draft Document for Review November 14, 2006 3:49 pm
Note: Before using this information and the product it supports, read the information in “Notices” on
page xxi.
This edition applies to the IBM System Storage DS8000 series Turbo Models 931, 932, and 9B2 as
announced in August 2006.
This document created or updated on November 14, 2006.
iii
6786edno.fm Draft Document for Review November 14, 2006 3:49 pm
Contents
Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii
The team that wrote this redbook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii
Special thanks to: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv
Become a published author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvi
Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvi
Contents vii
6786TOC.fm Draft Document for Review November 14, 2006 3:49 pm
Contents ix
6786TOC.fm Draft Document for Review November 14, 2006 3:49 pm
Contents xi
6786TOC.fm Draft Document for Review November 14, 2006 3:49 pm
Contents xiii
6786TOC.fm Draft Document for Review November 14, 2006 3:49 pm
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
Figures
Figures xvii
6786LOF.fm Draft Document for Review November 14, 2006 3:49 pm
Figures xix
6786LOF.fm Draft Document for Review November 14, 2006 3:49 pm
Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in other countries. Consult
your local IBM representative for information on the products and services currently available in your area. Any
reference to an IBM product, program, or service is not intended to state or imply that only that IBM product,
program, or service may be used. Any functionally equivalent product, program, or service that does not
infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to
evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this document. The
furnishing of this document does not give you any license to these patents. You can send license inquiries, in
writing, to:
IBM Director of Licensing, IBM Corporation, North Castle Drive, Armonk, NY 10504-1785 U.S.A.
The following paragraph does not apply to the United Kingdom or any other country where such
provisions are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION
PROVIDES THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR
IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of
express or implied warranties in certain transactions, therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically made
to the information herein; these changes will be incorporated in new editions of the publication. IBM may make
improvements and/or changes in the product(s) and/or the program(s) described in this publication at any time
without notice.
Any references in this information to non-IBM Web sites are provided for convenience only and do not in any
manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the
materials for this IBM product and use of those Web sites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring
any obligation to you.
Information concerning non-IBM products was obtained from the suppliers of those products, their published
announcements or other publicly available sources. IBM has not tested those products and cannot confirm the
accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the
capabilities of non-IBM products should be addressed to the suppliers of those products.
This information contains examples of data and reports used in daily business operations. To illustrate them
as completely as possible, the examples include the names of individuals, companies, brands, and products.
All of these names are fictitious and any similarity to the names and addresses used by an actual business
enterprise is entirely coincidental.
COPYRIGHT LICENSE:
This information contains sample application programs in source language, which illustrate programming
techniques on various operating platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM, for the purposes of developing, using, marketing or distributing application
programs conforming to the application programming interface for the operating platform for which the sample
programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore,
cannot guarantee or imply reliability, serviceability, or function of these programs.
Trademarks
The following terms are trademarks of the International Business Machines Corporation in the United States,
other countries, or both:
Redbooks (logo) ™ DS8000™ POWER™
eServer™ Enterprise Storage Server® Redbooks™
iSeries™ ESCON® Resource Link™
i5/OS® FlashCopy® RMF™
pSeries® FICON® RS/6000®
xSeries® Geographically Dispersed Parallel S/390®
z/OS® Sysplex™ System i™
z/VM® GDPS® System i5™
z/VSE™ HACMP™ System p™
zSeries® IBM® System x™
AIX 5L™ IMS™ System z™
AIX® Lotus® System Storage™
AS/400® Multiprise® System Storage DS™
BladeCenter® MVS™ Tivoli®
CICS® OS/2® TotalStorage®
DB2® OS/390® VM/ESA®
DFSMSdss™ OS/400® VSE/ESA™
DFSMShsm™ Parallel Sysplex® WebSphere®
DFSORT™ Power PC® 1-2-3®
DS4000™ PowerPC®
DS6000™ Predictive Failure Analysis®
Java and all Java-based trademarks are trademarks of Sun Microsystems, Inc. in the United States, other
countries, or both.
Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the
United States, other countries, or both.
Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel
SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its
subsidiaries in the United States and other countries.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Linux is a trademark of Linus Torvalds in the United States, other countries, or both.
Other company, product, or service names may be trademarks or service marks of others.
Preface
This IBM® Redbook describes the concepts, architecture and implementation of the IBM
System Storage™ DS8000 series of disk storage subsystems.
This book has reference information that will help you prepare the planning, installation, and
configuration of the DS8000 and also discusses the architecture and components. This book
will help you plan, design and implement a new installation, or migrate from an existing one. It
includes hints and tips derived from users experience for efficient installation and use.
The DS8000 series benefit by using IBM POWER5+™ processor technology with a dual
two-way processor complex implementation —in the DS8100 Turbo model 931; and with a
dual four-way processor complex implementation —in the DS8300 Turbo models 932 and
9B2. This increased power and extended connectivity, with up to 128 Fibre Channel/FICON
ports or 64 ESCON ports for host connections, makes it suitable for multiple server
environments in both the open systems and System z™ environments. Its switched Fibre
Channel architecture, dual processor complex implementation, high-availability design, and
the advanced point-in-time copy and remote mirror and copy functions that it incorporates
make the DS8000 suitable for mission-critical businesses.
To read about DS8000 point-in-time copy —FlashCopy, and the set of remote mirror and copy
products available with the DS8000 series —Metro Mirror, Global Copy, Global Mirror, z/OS
Global Mirror, and Metro/Global Mirror, you can refer to the redbooks: IBM System Storage
DS8000 Series: Copy Services in Open Environments, SG24-6788 and IBM System Storage
DS8000 Series: Copy Services with System z servers, SG24-6787.
Gustavo Castets is a Certified IBM IT Specialist and Project Leader working for the IBM
International Technical Support Organization, San Jose Center. While in San Jose, from 2001
to 2004, Gustavo co-authored twelve redbooks and taught IBM classes worldwide in the area
of disk and tape storage systems. Before joining the ITSO, Gustavo was based in Buenos
Aires where he worked in different technical sales support positions for more than 22 years.
Today, in addition to his work with the ITSO, Gustavo is working for IBM Global Delivery in
Argentina, as a Storage Specialist giving technical support to accounts from US and Europe.
Bert Dufrasne is a Certified Consulting IT Specialist and Project Leader for IBM
TotalStorage® and System Storage products at the International Technical Support
Organization, San Jose Center. He has worked at IBM in various IT areas. Before joining the
ITSO, he worked for IBM Global Services as an Application Architect. He holds a degree in
Electrical Engineering.
Stephen Baird is an IT Specialist with IBM Global Services. He joined IBM in 1999, working
in open systems server performance management and capacity planning. Since 2002, he has
worked in Storage Area Network and disk storage subsystem support and has gained
experience with Brocade, Cisco, and McData fiber channel switches and directors as well as
DS8000, DS4000, and ESS series hardware. He holds a degree in Mechanical Engineering
from the Massachusetts Institute of Technology, in Cambridge, MA.
Denise Brown is a Software Engineer at the IBM Open Systems Level Test Lab in Tucson,
Arizona. She has been working with IBM Storage for the past four years, with experience in
storage software and hardware in an open system environment, Her current areas of focus
include Copy Services solutions in Metro/Global Mirror and Incremental Re-synchronization
for the DS8000. She holds a degree in Engineering Mathematics.
Jana Jamsek is an IT specialist in IBM Slovenia. She works in Storage Advanced Technical
Support for Europe as a specialist for IBM Storage Systems and i5/OS systems. Jana has
eight years of experience in the System i and AS/400 area, and six years experience in
Storage. She holds a masters degree in computer science and a degree in mathematics from
University of Ljubljana, Slovenia. She was among the authors of the IBM Redpaper, The LTO
Ultrium Primer for IBM eServer iSeries Customers, the IBM Redbook, iSeries in Storage Area
Networks, the IBM Redbook, iSeries and IBM TotalStorage: A Guide to Implementing
External Disk on IBM eSerevr i5, and Redpaper, Multipath for IBM eServer iSeries.
Peter Klee is working as an IT Specialist with ATS EMEA, located in Mainz, Germany. He
has 10 years of experience in data centers. He worked for a large bank in Germany and was
responsible for the architecture and the implementation of the disk storage environment using
EMC Symmetrix, HDS Lightning and ESS Model 800. He joined IBM in 2003, where he was
working for Strategic Outsourcing. Since June 2004 he is working in the ATS System Storage
team in Mainz. His main focus is Copy Services in the open systems environment with
DS8000 and DS6000, especially Global Mirror and Metro/Global Mirror.
Markus Oscheka is an IT Specialist for Proof of Concepts and Benchmarks at the ATS
Customer Solutions team in Mainz, Germany. His areas of expertise include setup and
demonstration of IBM System Storage products and solutions in various environments
including AIX, Linux, Windows, HP-UX and Solaris. He has worked at IBM for five years. He
has performed many Proof of Concepts with Copy Services on DS6000/DS8000, as well as
Performance-Benchmarks with DS4000/DS6000/DS8000.
Ying Thia is an Advisory IT Specialist based in IBM Singapore, providing storage technical
support. She has 14 years of experience in the S/390 and storage environment. Before
joining the IBM Singapore Storage team, she worked in IBM Global Services where her
responsibilities include technical support and services delivery. Her areas of expertise include
IBM high-end disk and tape storage subsystems and disaster recovery solutions using the
capabilities and features of IBM storage products. She co-authored a previous redbook and
workshop for zSeries copy services.
Robert Tondini is a Senior IT specialist based in IBM Australia, providing storage technical
support. He has 12 years of experience in the S/390 and storage environment. His areas of
expertise include IBM high-end disk and tape storage subsystems and disaster recovery
solutions using the capabilities and features of IBM storage products. He co-authored several
redbooks and workshop for zSeries copy services.
The team: Gustavo, Robert, Wenzel, Jana, Peter, Markus, Denise, Werner, Ying, Stephen, Bertrand
We want to thank Michael Eggloff and Peter Klee for hosting us at the European Storage
Competency Center in Mainz, Germany. They were able to supply us with the needed
hardware, conference room, and all of the many things needed to run a successful residency.
Günter Schmitt, Uwe Schweikhard, Edgar Strubel (ATS - IBM Mainz) for their help in
reserving and preparing the equipment we used.
Monika Baier, Susanne Balzer, Marion Barlen, Marion Hartmann, Andrea Witkowski, Gertrud
Kramer - IBM Mainz, for their help with administrative tasks.
Selwyn Dickey, Timothy Klubertanz, Vess Natchev, James McCord and Chuck Stupca
IBM Rochester - System i Client Technology Center
Bob Bartfai, James Davison, Craig Gordon, Lee La Frese, Jennifer Mason, Alan McClure,
Rosemary McCutchen, Rachel Mikolajewski, Christopher o’Toole, Bill Raywinkle, Richard
Preface xxv
6786pref.fm Draft Document for Review November 14, 2006 3:49 pm
Ripberger, Henry Sautter, Jim Sedgwick, Gail Spear, David V Valverde, Steve Van Gundy,
Leann Vaterlaus, Sonny Williams, Steve Wilkins.
Brian Sherman
IBM Canada
Nick Clayton
IBM UK
Your efforts will help increase product acceptance and client satisfaction. As a bonus, you'll
develop a network of contacts in IBM development labs, and increase your productivity and
marketability.
Find out more about the residency program, browse the residency index, and apply online at:
ibm.com/redbooks/residencies.html
Comments welcome
Your comments are important to us!
We want our Redbooks to be as helpful as possible. Send us your comments about this or
other Redbooks in one of the following ways:
Use the online Contact us review redbook form found at:
ibm.com/redbooks
Send your comments in an email to:
redbook@us.ibm.com
Mail your comments to:
IBM Corporation, International Technical Support Organization
Dept. HYTD Mail Station P099
2455 South Road
Poughkeepsie, NY 12601-5400
Summary of changes
This section describes the technical changes made in this edition of the book and in previous
editions. This edition may also include minor corrections and editorial changes that are not
identified.
Summary of Changes
for SG24-6786-02
for IBM System Storage DS8000 Series: Architecture and Implementation
as created or updated on November 14, 2006.
New information
Turbo models 931/932/9B2 with POWER5+™ processor
500 GB - 7200 rpm Fibre Channel ATA (FATA) disk drives
4 Gb Fibre Channel/FICON host adapter connectivity option
IBM TotalStorage Productivity Center (TPC) for Replication V3.1 support
DS8000 series enhancements: End-to-End I/O Priorities; Cooperative Caching; Long
Busy Wait Host Tolerance; Audit Logging; z/OS Global Mirror suspend instead of long
busy option
Metro/Global Mirror reference information
Earthquake resistance kit
Added information on spare creation
System i™ related information
HyperPAV
242x and 239x Enterprise Choice length of warranty machine types
Added information on RPQ options for larger capacity in 932 Turbo Models
Changed information
The information, and some examples and screen captures presented in this book were
updated to reflect the latest available microcode bundle.
Expanded sections describing how PAV works
Expanded z/VM support information and z/VM support of PAVs
Merged contents from the publication IBM TotalStorage DS8000 series: Concepts and
Architecture, SG24-6452
Part 1
The policy-based management capabilities built into the IBM System Storage Open Software
Family, IBM DB2 Content Manager and IBM System Storage Archive Manager, can be used
with the DS8000 and are designed to help you automatically preserve critical data, while
preventing deletion of that data before its scheduled expiration.
Furthermore, the availability of Fibre Channel ATA (FATA) drives for the DS8000, gives the
option to retain frequently accessed or high-value data on Fibre Channel disk drives, and
archiving less valuable information on the less costly FATA disk drives —FATA drives are
discussed later in Section 4.4.2, “Disk enclosures” on page 50.
With the DS8300 Turbo Model 9B2, it is possible to create two storage system logical
partitions (LPARs) in a single DS8300. These LPARs divide the system resources equally and
can be used in completely separate storage environments.
The DS8000 series features several models in a new, higher-density footprint than the ESS
Model 800, providing configuration flexibility. The available models are discussed in
Chapter 2, “Model overview” on page 15.
Compared to the predecessor DS8000 series models (921/922/9A2) that featured the IBM
POWER5™ processor, the Turbo models with their POWER5+™ processor may enable up to
a 15% performance improvement in I/O operations per second in transaction processing
workload environments.
Internal fabric
DS8000 comes with a high bandwidth, fault tolerant internal interconnection, which is also
used in the IBM System p™ servers, called RIO-2 (Remote I/O). This technology can operate
at speeds up to 1 GHz and offers a 2 GB per second sustained bandwidth per link.
FATA drives
With the introduction of 500 GB (7200 rpm) Fibre Channel ATA (FATA) drives, the DS8000
capacity now scales up to 320 TB. These drives offers a cost effective option for lower priority
data. See 4.4.4, “Fiber Channel ATA (FATA) disk drives overview” on page 56 for more
information.
Host adapters
The DS8000 series offers enhanced connectivity with four-port Fibre Channel/FICON Host
Adapters. The DS8000 Turbo models offer 4Gb Fibre Channel/FICON host support designed
to offer up to 50% improvement in a single port MB/second throughput performance, helping
to enable cost savings with potential reduction in the number of host ports needed (compared
to 2Gb Fibre Channel and FICON support). The 2Gb Fibre Channel/FICON Host Adapters,
offered in long-wave and shortwave, auto-negotiate to either 2Gb, or 1Gb link speeds. The
4Gb Fibre Channel/FICON Host Adapters, also offered in long-wave and shortwave,
auto-negotiate to either 4Gb, 2Gb, or 1Gb link speeds. This flexibility enables the ability to
exploit the benefits offered by higher performance, 4Gb SAN-based solutions, while also
maintaining compatibility with existing 2Gb infrastructures. This feature continues to provide
individual ports on the adapter which can be configured with Fibre Channel Protocol (FCP) or
FICON. This can help protect your investment in Fibre Channel adapters, and increase your
ability to migrate to new servers. The DS8000 series also offers two-port ESCON adapters.
A DS8100 Turbo Model 931 can support up to a maximum of 16 host adapters. A DS8300
Turbo Model 932 or 9B2 can support up to a maximum of 32 host adapters, which provide up
to 128 Fibre Channel/FICON ports.
In the first frame, there is space for a maximum of 128 disk drive modules (DDMs) and each
expansion frame can contain 256 DDMs. With a maximum of 384 DDMs, the DS8100 Turbo
Model 931 using 500GB drives provides up to 192 TB of storage capacity. With a maximum of
640 DDMs, the DS8300 Turbo Models 932 and 9B2 using 500 GB drives provide up to
320 TB of storage capacity. In addition RPQs can be submitted for the Turbo Model 932 that
allow this model to scale up to 512TB of capacity —if fully configured with 500GB drives.
The operating system support for the DS8000 series is almost the same as for the previous
ESS Model 800; there are over 90 supported platforms. You can refer to the DS8000
interoperability matrix for the most current list of supported platforms. The DS8000
interoperability matrix can be found at:
http://www-03.ibm.com/servers/storage/disk/ds8000/pdf/ds8000-matrix.pdf
This rich support of heterogeneous environments and attachments, along with the flexibility to
easily partition the DS8000 series storage capacity among the attached environments, can
help support storage consolidation requirements and dynamic, changing environments.
For data protection and availability needs, the DS8000 provides Metro Mirror, Global Mirror,
Global Copy, and Metro/Global Mirror which are remote mirror and copy functions. These
functions are also available and are fully interoperable with the ESS 800 and 750 models and
the DS6800 series. These functions provide storage mirroring and copying over large
distances for disaster recovery or availability purposes.
Flashcopy
The primary objective of FlashCopy is to very quickly create a point-in-time copy of a source
volume on a target volume. The benefits of FlashCopy are that the point-in-time target copy is
immediately available for use for backups or testing and the source volume is immediately
released so that applications can continue processing, with minimal application downtime.
The target volume can be either a logical or physical copy of the data, with the latter copying
the data as a background process. In a z/OS environment FlashCopy can also operate at a
data set level.
Incremental FlashCopy
Incremental FlashCopy provides the capability to refresh a LUN or volume involved in a
FlashCopy relationship. When a subsequent FlashCopy is initiated, only the data required to
bring the target current to the source's newly established point-in-time is copied.
Consistency Groups
Consistency Groups can be used to maintain a consistent point-in-time copy across multiple
LUNs or volumes, or even multiple DS8000, DS6800, ESS 800, and ESS 750 systems.
Following is a summary of the remote mirror and copy options available with the DS8000.
Metro Mirror
Metro Mirror, previously called Peer-to-Peer Remote Copy (PPRC), provides a synchronous
mirror copy of LUNs or volumes at a remote site within 300km.
Global Copy
Global Copy, previously called Extended Distance Peer-to-Peer Remote Copy (PPRC-XD), is
a non-synchronous long distance copy option for data migration and backup.
Global Mirror
Global Mirror provides an asynchronous mirror copy of LUNs or volumes over virtually
unlimited distances. The distance is typically limited only by the capabilities of the network
and channel extension technology being used.
Metro/Global Mirror
Metro/Global Mirror is a three-site data replication solution for both the open systems and the
System z environments. Local site (site A) to intermediate site (site B) provides high
availability replication using synchronous Metro Mirror, and intermediate site (site B) to
remote site (site C) provides long distance disaster recovery replication using asynchronous
Global Mirror.
1.2.6 Interoperability
The DS8000 not only supports a broad range of server environments, but it can also perform
remote mirror and copy with the DS6000 and the ESS Models 750 and 800. This offers a
dramatically increased flexibility in developing mirroring and remote copy solutions, and also
the opportunity to deploy business continuity solutions at lower costs than have been
previously available.
For maintenance and service operations, the Storage Hardware Management Console
(S-HMC) is the focal point. The management console is a dedicated workstation that is
physically located (installed) inside the DS8000 subsystem and can automatically monitor the
state of your system, notifying you and IBM when service is required.
The S-HMC is also the interface for remote services (call home and call back) which can be
configured to meet customer requirements. It is possible to allow one or more of the following:
call on error (machine detected), connection for a few days (customer initiated), and remote
error investigation (service initiated). The remote connection between the management
console and the IBM service organization will be done via a virtual private network (VPN)
point-to-point connection over the internet or modem.
The DS8000 comes with an outstanding four year warranty, an industry first, on both
hardware and software.
Since the mirror solutions are compatible between the ESS and the DS8000 series, it is
possible to think about a setup for a disaster recovery solution with the high performance
DS8000 at the primary site and the ESS at the secondary site, where the same performance
is not required.
Obviously the DS8000 can deliver a higher throughput and scales higher than the DS6000,
but not all customers need this high throughput and capacity. You can choose the system that
fits your needs. Both systems support the same SAN infrastructure and the same host
systems.
So it is very easy to have a mixed environment with DS8000 and DS6000 disk storage
subsystems with common skills and management functions thus optimizing the cost
effectiveness of your storage solution.
Logical partitioning available with the DS8300 Turbo Model 9B2 is not available on the
DS6000.
With all these components, the DS8000 is positioned at the top of the high performance
category.
SDD is provided with the DS8000 series at no additional charge. Fibre Channel (SCSI-FCP)
attachment configurations are supported in the AIX, HP-UX, Linux, Windows, Novell
NetWare, and Sun Solaris environments.
For more information on SDD see Section 15.1.4, “Multipathing support — Subsystem Device
Driver (SDD)” on page 290.
Chapter 10, “Performance” on page 167, gives you more information about the performance
aspects of the DS8000 family.
Refer to IBM System Storage DS8000 Series: Copy Services with System z servers,
SG24-6787 for a detailed discussion of z/OS Global Mirror and related enhancements.
The DS8000 series currently has three Turbo models available: the DS8100 Turbo Model 931,
and the DS8300 Turbo Models 932 and 9B2. The difference in models is in the processors
and in the capability of storage system LPARs. The predecessors of the DS8000 series Turbo
models were the DS8000 series Models 921, 922, and 9A2.
The last position of the three characters 9xx refers to either a dual two-way processor
complex model (9x1), or a dual four-way processor complex model (9x2), or an expansion
frame with no processors (9xE). The middle position of the three characters means LPAR or
non-LPAR model —93x are non-LPAR models, and 9Bx is an LPAR model). You can also
order expansion frames with the base frame: 92E for non-LPAR models and 9AE for LPAR
models.
Figure 2-1 summarizes the naming rules that apply to the Turbo models 931, 932, 9B2, as
well as to the predecessor models 921, 922, and 9A2 —for these older models the x takes the
values 2 and A.
For example,
931 : Non-LPAR / 2-way base frame
9AE : LPAR / Expansion frame
Figure 2-1 Naming convention for Turbo models 931, 932, 9B2 and older 921, 922 and 9A2
Table 2-1 summarizes the possible machine types and models combinations.
2107 931 932 9B2 92E and 9AE Turbo models Same models
share same
242x 931 932 9B2 92E and 9AE Turbo models characteristics
Enterprise Choice and functionality
Note: The 242x machine type numbers used for ordering purposes have changed in order
to support the multiple warranty options, but the DS8000 Turbo Models are otherwise the
same systems announced in August 2006.
In the discussions throughout this book we almost make no references to the machine
types but to the models, because identical models share identical characteristics
independently of the machine type designation.
Figure 2-2 DS8100 base frame with covers removed (left) and with Model 92E (right)
Figure 2-3 on page 19 shows the maximum configuration of a DS8100 with the 931 base
frame plus a 92E expansion frame and provides the front view of the basic structure and
placement of the hardware components within both frames.
Figure 2-3 Maximum DS8100 configuration — 931 base unit and 92E expansion
The Model 9B2 also has the capability of providing two storage images through two storage
system LPARs within the same physical storage unit. Both storage images split the processor,
cache, and adapter resources in a 50:50 manner.
only two ESCON ports —and both must be ESCON ports. A DS8000 can have both
ESCON adapters and Fibre Channel/FICON adapters at the same time
Figure 2-4 displays a DS8300 Turbo Model 932 from the rear. On the left is the rear view with
closed covers. The right shows the rear view with no covers while the middle view is another
rear view but only with one cover off. This view shows the standard 19 inch rack mounted
hardware including disk drives, processor complexes, and I/O enclosures.
Figure 2-4 DS8300 Turbo Model 932/9B2 base frame rear views — with and without covers
The DS8300 Turbo models can connect to one or two expansion frames. This provides the
following configuration alternatives:
With one expansion frame the storage capacity and number of adapters of the DS8300
models can expand to:
– Up to 384 DDMs in total —as for the DS8100. This is a maximum disk storage capacity
of 115.2 TB with 300 GB FC DDMs and 192 TB with 500 GB FATA DDMs.
– Up to 32 host adapters (HAs). This can be an intermix of Fibre Channel/FICON
(four-port) and ESCON (two-port) adapters.
With two expansion frames the disk capacity of the DS8300 models expands to:
– Up to 640 DDMs in total for a maximum disk storage capacity of 192 TB when utilizing
300 GB FC DDMs and 320 TB with 500 GB FATA DDMs.
Figure 2-5 on page 21 shows a maximum configuration for a DS8300 Turbo Model with two
expansion frames —base Model 932 connects to Model 92E expansions and base Model
9B2 connects to Model 9AE expansions. This figure also shows the basic hardware
components and how they are distributed across all three frames.
Base Model 932 or 9B2 Expansion Model 92E or 9AE Expansion Model 92E or 9AE
128 Drives 256 Drives 256 Drives
Figure 2-5 DS8300 maximum configuration — base and two expansion frames
Note the additional I/O enclosures in the first expansion frame, which is the middle frame in
Figure 2-5. Each expansion frame has twice as many DDMs as the base frame, so with 128
DDMs in the base frame and 2 x 256 DDMs in the expansion frames, a total of 640 DDMs is
possible.
2.1.5 DS8300 Turbo Model 932 with four expansion frames — 512 TB
In addition to the standard configurations you can order, with the DS8300 Turbo Model 932
you have the possibility of submitting an RPQ (Request for Price Quotation) for a
configuration of base unit plus up to four expansion frames. With the fourth expansion frame
attached, the capacity of the DS8300 Turbo Model 932 scales up to 1024 DDMs in total. This
is a maximum disk storage capacity of 512TB —when using 500 GB FATA DDMs.
Figure 2-6 on page 22 shows a picture of a DS8300 Turbo Model 932 with the maximum
configuration of 1024 disk drives.
Figure 2-6 DS8300 Turbo Model 932 — maximum configuration with 1024 disk drives
Table 2-2 DS8000 series model comparison — 921, 922 and 9A2, and Turbo 931. 932 and 9B2.
Base Images Expansion Processor DDMs Processor Max. Host
Model Model type up to Memory Adapters
Depending on the DDM sizes, which can be different within a 9x1or 9x2 and the number of
DDMs, the total capacity is calculated accordingly.
Each Fibre Channel/FICON Host Adapter has four Fibre Channel ports, providing up to 128
Fibre Channel ports for a maximum configuration. Each ESCON Host Adapter has two ports,
therefore, the maximum number of ESCON ports possible is 64. There can be an intermix of
Fibre Channel/FICON and ESCON adapters, up to the maximum of 32 adapters.
Therefore, you can select a physical capacity from 1.1 TB (73 GB x 16 DDMs) up to 320 TB
(500 GB x 640 FATA DDMs). We summarize these characteristics in Table 2-3 on page 24.
Table 2-3 Capacity comparison — device adapters, DDMs, and storage capacity
2-way (Base 2-way (with 4-way (Base 4-way (w/ one 4-way (w/ two
frame only) Expansion frame only) Expansion Expansion
frame) frame) frames)
Device adapters 2 to 8 2 to 8 2 to 8 2 to 16 2 to 16
(1 - 4 Fibre Loops)
Physical capacity 1.1 to 64TB 1.1 to 192 TB 1.1 to 64TB 1.1 to 192 TB 1.1 to 320 TB
Adding DDMs
A significant benefit of the DS8000 series is the ability to add DDMs without disruption for
maintenance. IBM offers capacity on demand solutions that are designed to meet the
changing storage needs of rapidly growing e-business. The Standby Capacity on Demand
(CoD) offering is designed to provide you with the ability to tap into additional storage and is
particularly attractive if you have rapid or unpredictable storage growth. Up to four Standby
CoD disk drive sets (64 disk drives) can be factory or field installed into your system. To
activate, you simply logically configure the disk drives for use —a non-disruptive activity that
does not require intervention from IBM. Upon activation of any portion of a Standby CoD disk
drive set, you must place an order with IBM to initiate billing for the activated set. At that time,
you can also order replacement Standby CoD disk drive sets. For more information on the
Standby CoD offering refer to the DS8000 series announcement letter —IBM announcement
letters can be found at http://www.ibm.com/products.
When the first expansion frame is attached to a 9x2 base frame, you do need a disruptive
maintenance —the first expansion enclosure for the Models 9x2 has I/O enclosures and
these I/O enclosures must be connected into the existing RIO-G loops. If you install the base
frame and the first expansion frame for the Model 9x2 at the beginning, you don’t need a
disruptive upgrade to add DDMs. The expansion enclosure for the DS8100 model and the
second expansion enclosure for the DS8300 models has no I/O enclosure, therefore you can
attach them to the existing frame without disruption.
Hosts
Data Cache
and
I/O
Persistent
Enclosure
Memory
2-way
processor RIO-G
Disks
Disks
Figure 2-8 4-way model components
These two figures describe the main components of the I/O controller for the DS8000 series.
The main components include the I/O processors, data cache, internal I/O bus (RIO-G loop),
host adapters, and device adapters. Figure 2-7 is a 2-way model and Figure 2-8 is a 4-way
model. You can easily see that, if you upgrade from the 2-way model to the 4-way model, the
number of main components doubles within a storage unit.
DS8000 model conversions are disruptive. In addition, data may not be preserved during the
conversion. Data migration or backup/restore is your responsibility. Fee-based data migration
services are available from Global Services.
The IBM POWER5+ processor is an optional feature for the DS8100 Model 921 and DS8300
Models 922 and 9A2. The installation of this feature is disruptive. The following upgrade paths
to Turbo models 931, 932, and 9B2 equivalences, are available for the older models 921, 922,
and 9A2:
Model 921 upgrade to “Turbo Model 931 equivalence” —add Processor Upgrade feature
and complete a Processor Memory feature conversion.
Model 921 upgrade to “Turbo Model 932 equivalence” —first a Model 921 to 922 model
conversion, then add Processor Upgrade feature and complete Processor Memory feature
conversion.
Model 921 upgrade to “Turbo Model 9B2 equivalence” —first a Model 921 to 9A2 model
conversion, then add Processor Memory feature.
Model 922 upgrade to “Turbo Model 932 equivalence” —add Processor Upgrade feature
and complete a Processor Memory feature conversion.
Model 922 upgrade to “Turbo Model 9B2 equivalence” —first a Model 922 to 9A2 model
conversion, then add Processor Upgrade feature.
Model 9A2 upgrade to “Turbo Model 9B2 equivalence” —add Processor Upgrade feature
and complete Processor Memory feature conversion.
Model 9A2 upgrade to “Turbo Model 932 equivalence” —first a Model 9A2 to 922 model
conversion, then add Processor Upgrade feature.
For more information on model conversions and upgrade paths refer to the DS8000 series
Turbo models announcement letter. IBM announcement letters can be found at:
http://www.ibm.com/products.
LPAR is one component of the POWER5+ system technology that is part of the IBM
Virtualization Engine. Using IBM Virtualization Engine technology, selected models of the
DS8000 series can be used as a single, large storage system, or can be used as multiple
storage systems with logical partitioning (LPAR) capabilities. IBM LPAR technology, which is
unique in the storage industry, allows the resources of the storage system to be allocated into
separate logical storage system partitions, each of which is totally independent and isolated.
Virtualization Engine (VE) delivers the capabilities to simplify the infrastructure by allowing the
management of heterogeneous partitions/servers on a single system.
Partitions
When a multi-processor computer is subdivided into multiple, independent operating system
images, those independent operating environments are called partitions. The resources on
the system are allocated to specific partitions.
Resources
Resources are defined as a system’s processors, memory, and I/O slots. I/O slots can be
populated by different adapters, such as Ethernet, SCSI, Fibre Channel or other device
controllers. A disk is allocated to a partition by assigning it the I/O slot that contains the disk’s
controller.
are configuration rules, the granularity of the units of resources that can be allocated to
partitions is very flexible. It is possible to add just one resource, independently of the others
LPAR differs from physical partitioning in the way resources are grouped to form a partition.
Logical partitions do not need to conform to the physical boundaries of the hardware used to
build the server. LPAR adds more flexibility to select components from the entire pool of
available system resources.
Logical Partition
I/O I/O I/O I/O I/O I/O I/O I/O I/O I/O I/O I/O
Memory
Processor Processor
complex 0 complex 1
Storage
LPAR01 Facility LPAR11
Image 1
Storage
LPAR02 Facility LPAR12
Image 2
LPARxy
x=Processor complex number
y=Storage facility number
Figure 3-2 DS8300 Model 9B2 - LPAR and storage facility image (SFI)
The DS8300 series incorporates two four-way POWER5+ server processors; see Figure 3-2.
We call each of these a processor complex. Each processor complex on the DS8300 is
divided into two processor LPARs (a set of resources on a processor complex that support
the execution of an operating system). The storage facility image (SFI) is built from a pair of
processor LPARs, one on each processor complex.
Figure 3-2 shows that LPAR01 from processor complex 0 and LPAR11 from processor
complex 1 form storage facility image 1 (SFI 1). LPAR02 from processor complex 0 and
LPAR12 from processor complex 1 form the second storage facility image (SFI 2).
Storage facility images (SFIs) are also referred to as storage system LPARs —and
sometimes more briefly as storage LPARs.
In the DS8300 Model 9B2 the split of the resources among the two possible system storage
LPARs has a 50/50 ratio as depicted in Figure 3-3.
Storage Storage
Facility RIO-G I/O drawers RIO-G
Facility
Image 1 Image 1
(LPAR01) (LPAR11)
Storage Storage
RIO-G I/O drawers RIO-G
Facility Facility
Image 2 Image 2
(LPAR02) (LPAR12)
Storage
enclosures
Processor complex 0 HA HA DA HA HA DA
Processor complex 1
2 Processors
Processors I/O drawer 2 Processors
Processors
RIO-G interface 0 RIO-G interface
CD/DVD
CD/DVD
S-HMC
C C' D D'
Ethernet-Port Ethernet-Port
Ethernet-Port
boot data data boot data data boot data data boot data data
HA HA DA HA HA DA
Memory Memory
SCSI controller
I/O drawer SCSI controller
RIO-G interface
3 RIO-G interface
2 Processors 2 Processors
RIO-G interface RIO-G interface
HA HA DA HA HA DA
I/O drawer
2
Figure 3-4 shows the split of all available resources between the two SFIs —each SFI has
50% of all available resources.
I/O resources
For one SFI, the following hardware resources are required:
2 SCSI controllers with 2 disk drives each
2 Ethernet ports (to communicate with the S-HMC)
1 Thin Device Media Bay (for example, CD or DVD; can be shared between the LPARs)
Each SFI will have two physical disk drives in each processor complex. Each disk drive will
contain three logical volumes, the boot volume and two logical volumes for the memory save
dump function. These three logical volumes are then mirrored across the two physical disk
drives for each LPAR. In Figure 3-4, for example, the disks A/A' are mirrors. For the DS8300
Model 9B2, there will be four drives total in one physical processor complex.
Therefore, every LPAR has two processors and so every SFI has four processors.
The memory limit depends on the total amount of available memory in the whole system.
Currently there are the following memory allocations per storage facility available:
32 GB (16 GB per processor complex, 16 GB per SFI)
64 GB (32 GB per processor complex, 32 GB per SFI)
128 GB (64 GB per processor complex, 64 GB per SFI)
256 GB (128 GB per processor complex, 128 GB per SFI)
As a result of the strict separation of the two images, the following configuration options exist:
Each SFI is assigned to one dedicated RIO-G loop; if an image is offline, its RIO-G loop is
not available.
All I/O enclosures, host adapters, and device adapters on a given RIO-G loop are
dedicated to the image that owns the loop.
Disk enclosures and storage devices behind a given device adapter pair are dedicated to
the image that owns the RIO-G loop.
Configuring of capacity to an image is managed through the placement of disk enclosures
on a specific DA pair dedicated to this image.
The Turbo Model 9B2 (base frame) has, divided evenly between the SFIs:
– 32 to 128 DDMs, in increments of 16
– System memory of 32, 64, 128, 256 GB
– Four I/O bays. Each bay contains:
• Up to 4 host adapters (divided between SFIs at the bay level)
• Up to 2 device adapters (divided between SFIs at the bay level)
– S-HMC, keyboard/display, and 2 Ethernet switches (shared between SFIs)
The first expansion frame Model 9AE has, divided evenly between the SFIs:
– An additional four I/O bays. Each bay contains:
• Up to 4 host adapters (divided between SFIs at the bay level)
• Up to 2 device adapters (divided between SFIs at the bay level)
– An additional 256 DDMs
The second expansion frame Model 9AE has, divided evenly between the SFIs:
– An additional 256 DDMs
A fully configured DS8300 has one base frame and two expansion frames. The first
expansion frame has additional I/O drawers and disk drive modules (DDMs), while the
second expansion frame contains only additional DDMs.
Figure 3-5 on page 34 provides an example of how a fully populated DS8300 might be
configured. The disk enclosures are assigned to SFI 1 (yellow) or SFI 2 (green). When
ordering additional disk capacity, it can be allocated to either SFI, but the cabling is
pre-determined so in this example disks added to the empty pair of disk enclosures would be
allocated to SFI 2.
Storage Storage
Facility
Processor
Facility
Storage Empty storage
complex 0
Image 1 Image 2 enclosure enclosure
Storage Storage
Storage Storage
Facility
Processor
Facility
enclosure enclosure
Image 1 complex 1 Image 2
Storage Storage
enclosure enclosure
In a partitioned environment, the POWER Hypervisor is loaded into the first Physical Memory
Block (PMB) at the physical address zero and reserves the PMB. From then on, it is not
possible for an LPAR to access directly the physical memory. Every memory access is
controlled by the POWER Hypervisor as shown in Figure 3-6 on page 35.
Each partition has its own exclusive page table, also controlled by the POWER Hypervisor,
which the processors use to transparently convert a program's virtual address into the
physical address where that page has been mapped into physical memory.
In a partitioned environment, the operating system uses hypervisor services to manage the
translation control entry (TCE) tables. The operating system communicates the desired I/O
bus address to logical mapping, and the hypervisor translates that into the I/O bus address to
physical mapping within the specific TCE table. The hypervisor needs a dedicated memory
region for the TCE tables to translate the I/O address to the partition memory address, then
the hypervisor can perform direct memory access (DMA) transfers to the PCI adapters.
Hypervisor-
Hypervisor- Controlled
Controlled Physical TCE Tables
Partition 1 Page Tables Memory For DMA Partition 1
Proc I/O Load/Store I/O Slot
Virtual Addresses
N
Bus Addresses
Proc I/O Slot
N
{
N
I/O Slot
Real Addresses Addr N
Partition 2
I/O Load/Store
Proc Virtual Addresses
0 Partition 2
Bus Addresses
Proc I/O Slot
0
Proc
0
Real Addresses
{ Addr 0
I/O Slot
Remote
mirror
9
Storage Facility Image 1 Storage Facility Image 2
FlashCopy FlashCopy
Secondary Source Target
FlashCopy
Source
8 FlashCopy
Target
9Remote mirroring within a Storage Facility Image or across Storage Facility Images
9FlashCopy within a Storage Facility Image
Worldwide Storage Kickoff Meeting | IBM Confidential
Figure 3-7 DS8300 storage facility images and Copy Services
Figure 3-7 summarizes the basic considerations for Copy Services when used with a
partitioned DS8300.
FlashCopy
The DS8000 series fully supports the FlashCopy V2 capabilities including cross LSS support.
However, a source volume of a FlashCopy located in one SFI cannot have a target volume in
the second SFI, as illustrated in Figure 3-7.
For more information about Copy Services refer to Chapter 7, “Copy Services” on page 107.
The hardware-based LPAR implementation ensures data integrity. The fact that you can
create dual, independent, completely segregated virtual storage systems helps you to
optimize the utilization of your investment, and helps to segregate workloads and protect
them from one another.
The following are examples of possible scenarios where SFIs would be useful:
Separating workloads
An environment can be split by operating system, application, organizational boundaries
or production readiness; for example, z/OS hosts on one SFI and open hosts on the other
or production on one and test or development on the other.
Dedicating resources
As a service provider you could provide dedicated resources to each customer, thereby
satisfying security and service level agreements, while having the environment all
contained on one physical DS8300.
Production and data mining
For database purposes, imagine a scenario where the production database is running in
the first SFI and a copy of the production database is running in the second SFI. Analysis
and data mining can be performed on it without interfering with the production database.
Business continuance (secondary) within the same physical array
You can use the two partitions to test Copy Services solutions or you can use them for
multiple copy scenarios in a production environment.
Information Lifecycle Management (ILM) partition with fewer resources, slower DDMs
One SFI can utilize, for example, only fast disk drive modules to ensure high performance
for the production environment, and the other SFI can use fewer and slower DDMs to
ensure Information Lifecycle Management at a lower cost.
Figure 3-8 on page 37 depicts one example of SFIs use in the DS8300.
System 1 System 2
Open System System z
LUN0
LUN1 3390-3
LUN2 3390-3
This example shows a DS8300 with a total physical capacity of 30 TB. In this case, a
minimum Operating Environment License (OEL) is required to cover the 30 TB capacity. The
DS8300 is split into two images. SFI 1 is used for an open system environment and utilizes 20
TB of fixed block data (FB). SFI 2 is used for a System z environment and uses 10 TB of
count key data (CKD).
3.6 Summary
The DS8000 series started delivering the first use of the POWER5 processor IBM
Virtualization Engine logical partitioning capability originally with the Model 9A2 and now
enhanced with the POWER5+ processor of the Turbo Model 9B2. This storage system LPAR
technology is designed to enable the creation of two completely separate storage systems.
The SFIs can be used for production, test, or other unique storage environments, and they
operate within a single physical enclosure. Each SFI can be established to support the
specific performance requirements of a different, heterogeneous workload. The DS8000
series robust partitioning implementation helps to isolate and protect the SFIs. These storage
system LPAR capabilities are designed to help simplify systems by maximizing management
efficiency, cost effectiveness, and flexibility.
4.1 Frames
The DS8000 is designed for modular expansion. From a high-level view there appear to be
three types of frames available for the DS8000. However, on closer inspection, the frames
themselves are almost identical. The only variations are what combinations of processors, I/O
enclosures, batteries, and disks the frames contain.
Figure 4-1 is an attempt to show some of the frame variations that are possible with the
DS8000 series. The left-hand frame is a base frame that contains the processors (System p5
servers). The center frame is an expansion frame that contains additional I/O enclosures but
no additional processors. The right-hand frame is an expansion frame that contains just disk
(and no processors, I/O enclosures, or batteries). Each frame contains a frame power area
with power supplies and other power-related hardware.
Battery
Battery
Backup unit
I/O I/O Backup unit I/O I/O
Battery Enclosure 1 enclosure 0 Battery Enclosure 5 enclosure 4
Backup unit backup unit
I/O I/O I/O I/O
Battery Battery
Backup unit Enclosure 3 enclosure 2 Enclosure 7 enclosure 6
Backup unit
The base frame can contain up to eight disk enclosures, each can contain up to 16 disk
drives. In a maximum configuration, the base frame can hold 128 disk drives. Above the disk
enclosures are cooling fans located in a cooling plenum.
Between the disk enclosures and the processor complexes are two Ethernet switches, a
Storage Hardware Management Console (an S-HMC) and a keyboard/display module.
The base frame contains two processor complexes. These System p5 servers contain the
processor and memory that drive all functions within the DS8000. In the ESS we referred to
them as clusters, but this term is no longer relevant. We now have the ability to logically
partition each processor complex into two LPARs, each of which is the equivalent of an ESS
cluster.
Finally, the base frame contains four I/O enclosures. These I/O enclosures provide
connectivity between the adapters and the processors. The adapters contained in the I/O
enclosures can be either device or host adapters (DAs or HAs). The communication path
used for adapter to processor complex communication is the RIO-G loop. This loop not only
joins the I/O enclosures to the processor complexes, it also allows the processor complexes
to communicate with each other.
Each expansion frame can hold up to 16 disk enclosures which contain the disk drives. They
are described as 16-packs because each enclosure can hold 16 disks. In a maximum
configuration, an expansion frame can hold 256 disk drives. Above the disk enclosures are
cooling fans located in a cooling plenum.
An expansion frame can contain I/O enclosures and adapters if it is the first expansion frame
that is attached to a 9x2 model. The second expansion frame in a 9x2 model configuration
cannot have I/O enclosures and adapters, nor can any expansion frame that is attached to a
model 9x1. If the expansion frame contains I/O enclosures, the enclosures provide
connectivity between the adapters and the processors. The adapters contained in the I/O
enclosures can be either device or host adapters. The available expansion frame models are:
Model 92E expansion unit —attaches to the DS8000 series Turbo models 931 and 932,
and to the older DS8000 series 921 and 922 models
Model 9AE expansion unit —attaches to the DS8000 series Turbo Model 9B2 and to the
older DS8000 series 9A2 model.
There is also an EPO switch on each operator panel. This switch is only for emergencies.
Tripping the EPO switch will bypass all power sequencing control and result in immediate
removal of system power. A small cover must be lifted to operate it. Do not trip this switch
unless the DS8000 is creating a safety hazard or is placing human life at risk.
Line cord
indicators Fault indicator
You will note that there is not a power on/off switch on the operator panel. This is because
power sequencing is managed via the S-HMC. This is to ensure that all data in non-volatile
storage (known as modified data) is de-staged properly to disk prior to power down. It is thus
not possible to shut down or power off the DS8000 from the operator panel (except in an
emergency, with the EPO switch mentioned previously).
4.2 Architecture
Now that we have described the frames themselves, we use the rest of this chapter to explore
the technical details of each of the components. The architecture that connects these
components is pictured in Figure 4-3 on page 43.
In effect, the DS8000 consists of two processor complexes. Each processor complex has
access to multiple host adapters to connect to Fibre Channel, FICON and ESCON hosts.
Each DS8000 can have up to 32 host adapters. To access the disk subsystem, each complex
uses several four-port Fibre Channel arbitrated loop (FC-AL) device adapters. A DS8000 can
have up to sixteen of these adapters arranged into eight pairs. Each adapter connects the
complex to two separate switched Fibre Channel networks. Each switched network attaches
disk enclosures that each contain up to 16 disks. Each enclosure contains two 20-port Fibre
Channel switches. Of these 20 ports, 16 are used to attach to the 16 disks in the enclosure
and the remaining four are used to either interconnect with other enclosures or to the device
adapters. Each disk is attached to both switches. Whenever the device adapter connects to a
disk, it uses a switched connection to transfer data. This means that all data travels via the
shortest possible path.
The attached hosts interact with software which is running on the complexes to access data
on logical volumes. Each complex will host at least one instance of this software (which is
called a server), which runs in a logical partition (an LPAR). The servers manage all read and
write requests to the logical volumes on the disk arrays. During write requests, the servers
use fast-write, in which the data is written to volatile memory on one complex and persistent
memory on the other complex. The server then reports the write as complete before it has
been written to disk. This provides much faster write performance. Persistent memory is also
called NVS or non-volatile storage.
RIO-G
N-way First RIO-G loop N-way
SMP SMP
When a host performs a read operation, the servers fetch the data from the disk arrays via the
high performance switched disk architecture. The data is then cached in volatile memory in
case it is required again. The servers attempt to anticipate future reads by an algorithm
known as SARC (Sequential prefetching in Adaptive Replacement Cache). Data is held in
cache as long as possible using this smart algorithm. If a cache hit occurs where requested
data is already in cache, then the host does not have to wait for it to be fetched from the disks.
Both the device and host adapters operate on a high bandwidth fault-tolerant interconnect
known as the RIO-G. The RIO-G design allows the sharing of host adapters between servers
and offers exceptional performance and reliability.
Figure 4-3 uses colors as indicators of how the DS8000 hardware is shared between the
servers (the cross hatched color is green and the lighter color is yellow). On the left side, the
green server is running on the left-hand processor complex. The green server uses the N-way
SMP of the complex to perform its operations. It records its write data and caches its read
data in the volatile memory of the left-hand complex. For fast-write data it has a persistent
memory area on the right-hand processor complex. To access the disk arrays under its
management (the disks also being pictured in green), it has its own device adapter (again in
green). The yellow server on the right operates in an identical fashion. The host adapters (in
dark red) are deliberately not colored green or yellow because they are shared between both
servers.
The design decision to use SMP memory as I/O cache is a key element of IBM’s storage
architecture. Although a separate I/O cache could provide fast access, it cannot match the
access speed of the SMP main memory. The decision to use the SMP main memory as the
cache proved itself in three generations of IBM TotalStorage Enterprise Storage Server
(ESS). The performance roughly doubled with each generation. This performance
improvement can be traced to the capabilities of the completely integrated SMP, the
processor speeds, the L1/L2 cache sizes and speeds, the memory bandwidth and response
time, and the PCI bus performance.
With the DS8000, the cache access has been accelerated further by making the Non-Volatile
Storage a part of the SMP memory.
All memory installed on any processor complex is accessible to all processors in that
complex. The addresses assigned to the memory are common across all processors in the
same complex. On the other hand, using the main memory of the SMP as the cache, leads to
a partitioned cache. Each processor has access to the processor complex’s main memory but
not to that of the other complex. You should keep this in mind with respect to load balancing
between processor complexes.
The DS8000 (and the DS6000) use the Sequential Prefetching in Adaptive Replacement
Cache (SARC) algorithm, developed by IBM Storage Development in partnership with IBM
Research. It is a self-tuning, self-optimizing solution for a wide range of workloads with a
varying mix of sequential and random I/O streams. SARC is inspired by the Adaptive
Replacement Cache (ARC) algorithm and inherits many features from it. For a detailed
description of ARC see N. Megiddo and D. S. Modha, “Outperforming LRU with an adaptive
replacement cache algorithm,” IEEE Computer, vol. 37, no. 4, pp. 58–65, 2004.
The DS8000 cache is organized in 4K byte pages called cache pages or slots. This unit of
allocation (which is smaller than the values used in other storage systems) ensures that small
I/Os do not waste cache memory.
The decision to copy some amount of data into the DS8000 cache can be triggered from two
policies: demand paging and prefetching. Demand paging means that eight disk blocks (a 4K
cache page) are brought in only on a cache miss. Demand paging is always active for all
volumes and ensures that I/O patterns with some locality find at least some recently used
data in the cache.
Prefetching means that data is copied into the cache speculatively even before it is
requested. To prefetch, a prediction of likely future data accesses is needed. Because
effective, sophisticated prediction schemes need extensive history of page accesses (which is
not feasible in real-life systems), SARC uses prefetching for sequential workloads. Sequential
access patterns naturally arise in video-on-demand, database scans, copy, backup and
recovery. The goal of sequential prefetching is to detect sequential access and effectively
pre-load the cache with data so as to minimize cache misses.
For prefetching, the cache management uses tracks. A track is a set of 128 disk blocks
(16 cache pages). To detect a sequential access pattern, counters are maintained with every
track to record if a track has been accessed together with its predecessor. Sequential
prefetching becomes active only when these counters suggest a sequential access pattern. In
this manner, the DS8000 monitors application read-I/O patterns and dynamically determines
whether it is optimal to stage into cache:
Just the page requested
That page requested plus remaining data on the disk track
An entire disk track (or a set of disk tracks) which has (have) not yet been requested
The decision of when and what to prefetch is essentially made on a per-application basis
(rather than a system-wide basis) to be sensitive to the different data reference patterns of
different applications that can be running concurrently.
To decide which pages are evicted when the cache is full, sequential and random
(non-sequential) data is separated into different lists; see Figure 4-4.
RANDOM SEQ
MRU MRU
Desired size
SEQ bottom
LRU
RANDOM bottom
LRU
Figure 4-4 Cache lists of the SARC algorithm for random and sequential data
A page which has been brought into the cache by simple demand paging is added to the
MRU (Most Recently Used) head of the RANDOM list. Without further I/O access, it goes
down to the LRU (Least Recently Used) bottom. A page which has been brought into the
cache by a sequential access or by sequential prefetching is added to the MRU head of the
SEQ list and then goes in that list. Additional rules control the migration of pages between the
lists so as to not keep the same pages in memory twice.
To follow workload changes, the algorithm trades cache space between the RANDOM and
SEQ lists dynamically and adaptively. This makes SARC scan-resistant, so that one-time
sequential requests do not pollute the whole cache. SARC maintains a desired size
parameter for the sequential list. The desired size is continually adapted in response to the
workload. Specifically, if the bottom portion of the SEQ list is found to be more valuable than
the bottom portion of the RANDOM list, then the desired size is increased; otherwise, the
desired size is decreased. The constant adaptation strives to make optimal use of limited
cache space and delivers greater throughput and faster response times for a given cache
size.
Additionally, the algorithm modifies dynamically not only the sizes of the two lists, but also the
rate at which the sizes are adapted. In a steady state, pages are evicted from the cache at the
rate of cache misses. A larger (respectively, a smaller) rate of misses effects a faster
(respectively, a slower) rate of adaptation.
Other implementation details take into account the relation of read and write (NVS) cache,
efficient de-staging, and the cooperation with Copy Services. In this manner, the DS8000
cache management goes far beyond the usual variants of the LRU/LFU (Least Recently Used
/ Least Frequently Used) approaches.
The DS8000 series features IBM POWER5 server technology. Depending on workload, the
maximum host I/O operations per second of the DS8100 9x1 models is up to three times the
maximum operations per second of the ESS Model 800. The maximum host I/O operations
per second of the DS8300 9x2 models is up to six times the maximum of the ESS Model 800.
For details on the server hardware used in the DS8000, refer to the redpaper IBM p5 570
Technical Overview and Introduction, REDP-9117, available at:
http://www.redbooks.ibm.com
true 16-way SMP combined system. How these features are implemented in the DS8000
might vary.
Figure 4-5 shows a front view and a rear view of the DS8000 series processor complex.
Power supply 1
PCI-X adapters in
Power supply 2 blind-swap carriers
Front view
DVD-rom drives
Operator panel
Processor cards
Power supply 1
Power supply 2
Rear view
PCI-X slots
RIO-G ports RIO-G ports
Figure 4-5 Processor complex
The System p5 servers in the DS8000 include two 3-pack front-accessible, hot-swap-capable
disk bays. The six disk bays of one System p5 processor complex can accommodate up to
880.8 GB of disk storage using the 146.8 GB Ultra320 SCSI disk drives. Two additional media
bays are used to accept optional slim-line media devices, such as DVD-ROM or DVD-RAM
drives. The System p5 processor complex also has I/O expansion capability using the RIO-G
interconnect. How these features are implemented in the DS8000 might vary.
Processor memory
The DS8100 models 9x1 offer up to 128 GB of processor memory and the DS8300 models
9x2 offer up to 256 GB of processor memory. Half of this will be located in each processor
complex. In addition, the Non-Volatile Storage (NVS) scales to the processor memory size
selected, which can also help optimize performance.
The SP performs predictive failure analysis based on any recoverable processor errors. The
SP can monitor the operation of the firmware during the boot process, and it can monitor the
operating system for loss of control. This enables the service processor to take appropriate
action.
The SPCN monitors environmentals such as power, fans, and temperature. Environmental
critical and non-critical conditions can generate Early Power-Off Warning (EPOW) events.
Critical events trigger appropriate signals from the hardware to the affected components to
prevent any data loss without operating system or firmware involvement. Non-critical
environmental events are also logged and reported.
4.3.1 RIO-G
The RIO-G ports are used for I/O expansion to external I/O drawers. RIO stands for remote
I/O. The RIO-G is evolved from earlier versions of the RIO interconnect.
Each RIO-G port can operate at 1 GHz in bidirectional mode and is capable of passing data in
each direction on each cycle of the port. It is designed as a high performance self-healing
interconnect. Each System p5 server in the DS8000 provides two external RIO-G ports, and
an adapter card adds two more. Two ports on each processor complex form a loop.
Figure 4-6 on page 49 illustrates how the RIO-G cabling is laid out in a DS8000 that has eight
I/O drawers. This would only occur if an expansion frame were installed. The DS8000 RIO-G
cabling will vary based on the model. A two-way DS8000 model will have one RIO-G loop. A
four-way DS8000 model will have two RIO-G loops. Each loop will support four disk
enclosures.
Processor Processor
Complex 0 Complex 1
Loop 0
RIO-G ports RIO-G ports
Loop 1
RIO-G
Redundant ports
power supplies
Slots: 1 2 3 4 5 6
Figure 4-7 I/O enclosures
The DAs are installed in pairs because each storage partition requires its own adapter to
connect to each disk enclosure for redundancy. This is why we refer to them as DA pairs.
dummy carriers. A dummy carrier looks very similar to a DDM in appearance but contains no
electronics. The enclosure is pictured in Figure 4-9.
Note: If a DDM is not present, its slot must be occupied by a dummy carrier. This is
because without a drive or a dummy, cooling air does not circulate correctly.
Each DDM is an industry standard FC-AL disk. Each disk plugs into the disk enclosure
backplane. The backplane is the electronic and physical backbone of the disk enclosure.
The full loop is required to participate in data transfer. Full discovery of the loop via LIP
(loop initialization protocol) is required before any data transfer. Loop stability can be
affected by DDM failures.
In the event of a disk failure, it can be difficult to identify the cause of a loop breakage,
leading to complex problem determination.
There is a performance dropoff when the number of devices in the loop increases.
To expand the loop it is normally necessary to partially open it. If mistakes are made, a
complete loop outage can result.
These problems are solved with the switched FC-AL implementation on the DS8000.
The DS8000 architecture employs dual redundant switched FC-AL access to each of the disk
enclosures. The key benefits of doing this are:
Two independent networks to access the disk enclosures.
Four access paths to each DDM.
Each device adapter port operates independently.
Double the bandwidth over traditional FC-AL loop implementations.
In Figure 4-11 each DDM is depicted as being attached to two separate Fibre Channel
switches. This means that with two device adapters, we have four effective data paths to each
disk, each path operating at 2Gbps. Note that this diagram shows one switched disk network
attached to each DA. Each DA can actually support two switched networks.
When a connection is made between the device adapter and a disk, the connection is a
switched connection that uses arbitrated loop protocol. This means that a mini-loop is created
between the device adapter and the disk. Figure 4-12 on page 53 depicts four simultaneous
and independent connections, one from each device adapter port.
Switched connections
15
Rear storage
enclosure N max=6
0 FC switch
Rear
8 or 16 DDMs per
enclosures Rear storage
15
enclosure
enclosure 2 0
2Gbs FC-AL link
4 FC-AL Ports
15
Rear storage
enclosure 1
0
Front storage
enclosure 1 15
enclosures enclosure 2 15
0
Front storage
enclosure N max=6
15
Expansion
Storage enclosures are added in pairs and disks are added in groups of 16. On the ESS
Model 800, the term 8-pack was used to describe an enclosure with eight disks in it. For the
DS8000, we use the term 16-pack, though this term really describes the 16 DDMs found in
one disk enclosure. It takes two orders of 16 DDMs to fully populate a disk enclosure pair
(front and rear).
To provide an example, if a machine had six disk enclosures total, it would have three at the
front and three at the rear. If all the enclosures were fully populated with disks, and an
additional order of 16 DDMs was purchased, then two new disk enclosures would be added,
one at the front and one at the rear. The switched networks do not need to be broken to add
these enclosures. They are simply added to the end of the loop, 8 DDMs would go in the front
enclosure and the remaining 8 ddms would go in the rear enclosure. If additional 16 DDMs
were ordered later, they would be used to fill up that pair of disk enclosures.
The intention is to only have four spares per DA pair, but this number may increase depending
on DDM intermix. Four DDMs of the largest capacity and at least two DDMs of the fastest
RPM are needed. If all DDMs are the same size and RPM, four spares are sufficient.
Server Server
0 Front enclosure 1
device Rear enclosure device
adapter adapter
3
loop 1 loop 1
4
There are two separate
switches in each enclosure.
Figure 4-14 DS8000 switched loop layout
To better understand AAL refer to Figure 4-14 and Figure 4-15. To make the diagrams clearer,
only 16 DDMs are shown, eight in each disk enclosure. When fully populated, there would be
16 DDMs in each enclosure.
Figure 4-14 is used to depict the device adapter pair layout. One DA pair creates two switched
loops. The front enclosures populate one loop while the rear enclosures populate the other
loop. Each enclosure places two switches onto each loop. Each enclosure can hold up to 16
DDMs. DDMs are purchased in groups of 16. Half of the new DDMs go into the front
enclosure and half go into the rear enclosure.
Having established the physical layout, the diagram is now changed to reflect the layout of the
array sites, as shown in Figure 4-15. Array site 0 in green (the darker disks) uses the four
left-hand DDMs in each enclosure. Array site 1 in yellow (the lighter disks), uses the four
right-hand DDMs in each enclosure. When an array is created on each array site, half of the
array is placed on each loop. A fully populated enclosure would have four array sites.
Server Server
0 1
Array site 0 Array site 1
device device
adapter adapter
3
loop 1 loop 1
4
AAL benefits
AAL is used to increase performance. When the device adapter writes a stripe of data to a
RAID-5 array, it sends half of the write to each switched loop. By splitting the workload in this
manner, each loop is worked evenly, which improves performance. If RAID-10 is used, two
RAID-0 arrays are created. Each loop hosts one RAID-0 array. When servicing read I/O, half
of the reads can be sent to each loop, again improving performance by balancing workload
across loops.
SATA disks use a single serial port interface on an ATA drive and offer a low-cost disk
technology to address less intensive storage operations. SATA drives for instance are used in
the IBM DS4000 Storage Server (to overcome the single port drive limitation, IBM uses a
MUX or interposer card to provide dual port access to the SATA drives).
Meant for capacity intensive, secondary, or near-line storage applications, SATA drive
reliability is similar to Fibre Channel drives when used within their recommended duty-cycle in
less I/O intensive applications.
In February 2002, a second ATA specification was launched called Serial ATA II. Second
generation SATA-2 disk drives have made significant improvements in speed and functionality
over the first generation SATA-1 by offering up to 3Gb/s speed and Native Tag Command
Queuing like Fibre Channel disks.
Designed to meet the architectural standards of enterprise-class storage systems FATA disk
drives provide high capacity at a low-cost alternative to FC disks without much sacrifice of
performance, availability or functionality. Also, when used within their recommended
duty-cycle reliability is comparable to that of the FC disks.
FATA disk drives are now available for both the IBM System Storage DS8000 series and
DS6000 series giving a lower cost alternative for large capacity, low workload environments.
Important: FATA is not the appropriate answer to every storage requirement. For many
enterprise applications, and certainly mission-critical and production applications, Fibre
Channel disks remain the best choice.
SATA and FATA disk drives are a cost efficient storage option for lower intensity storage
workloads. By providing the same dual port FC interface as Fibre Channel disks, FATA drives
offer higher availability and ensure compatibility and investment protection for existing
enterprise-class storage systems.
Note: The FATA drives offer a cost effective option for lower priority data such as various
fixed content, data archival, reference data, and near-line applications that require large
amounts of storage capacity for lighter workloads. These new drives are meant to comple-
ment, not compete with existing Fibre Channel drives as they are not intended for use in
applications that require drive utilization duty cycles greater than 20 percent.
Following is a summary of the key characteristics of the different disk drive types.
Fibre Channel
Intended for heavy workloads in multi-user environments
Highest performance, availability, reliability and functionality
Good Capacity: 36–300 GB
Very high activity
Greater than 80% duty-cycle
FATA
Intended for lower workloads in multi-user environments
High performance, availability and functionality
High reliability
More robust technology
– Extensive Command Queuing
High Capacity: 500 GB disk drives
Moderate activity
20-30% duty-cycle
SATA-1
Intended for lower workloads in multi-user environments
Good performance
Less availability and functionality than FATA or Fibre Channel disk drives
– Single port interface, no command queuing
High reliability
High capacity: 250–500 GB disk drives
Moderate activity
20-30% duty-cycle
SATA-2
Intended for lower workloads in multi-user environments
High performance, availability and functionality
High reliability
More robust technology
– Extensive Command Queuing
High Capacity: 500 GB disk drives
Moderate activity
20-30% duty-cycle
Without any doubt, the technical characteristics and performance of FC disks remain superior
to those of FATA disks. However, not all storage applications require these superior features.
When used for the appropriate enterprise applications, FATA disks offer a tremendous cost
advantage over FC. First, FATA drives are cheaper to manufacture and because of their lager
individual capacity, they are cheaper per gigabyte (GB) than FC disks. In large capacity
systems, the drives themselves account for the vast majority of the cost of the system. Using
FATA disks can substantially reduce the total cost of ownership (TCO) of the storage system.
Classes of storage
To better visualize where the benefits of FATA can best be obtained if implemented in a
networked storage environment, a positioning of the types or classes of storage and the
appropriate storage technology used at these levels helps in this observation.
Network
Basically, storage data can reside at three different locations within the networked storage
hierarchy. See Figure 16.
Particular data types are suitable for storage at the various levels:
Online (primary) storage
Best suited for applications that require constant instantaneous access to data, such as
databases and frequently accessed user data.
Primary storage stores business-critical information, data with the highest value and
importance. This data requires continuous availability and typically has high-performance
requirements. Business-critical data will be stored on Fibre Channel disk implemented in
enterprise-class storage solutions.
Near-line (secondary) storage
Used for applications that require quicker access compared with offline storage (such as
tape), but do not require the continuous, instantaneous access provided by online storage.
Secondary storage stores business-important information, but can, however, often tolerate
lower performance and potentially slightly less than 24/7 availability. It can also be used to
cache online storage for quicker backups to tape. Secondary storage represents a large
percentage of a company’s data and is an ideal fit for FATA technology.
Offline (archival) storage
Used for applications where infrequent serial access is required, such as backup for
long-term storage. For this type of storage, tape remains the most economical solution.
Data storage implementations best suited to use FATA technology reside at the “near-line” or
secondary location within the networked storage hierarchy and offer a cost-effective
alternative to FC disks at that location. Positioned between online storage and offline storage,
near-line storage or secondary storage is an optimal cost/performance solution for hosting
cached backups and fixed data storage.
Table 4-1 summarizes the general characteristics for primary, secondary, and archival storage
in traditional IT environments.
Fibre Channel drives were designed for the highest levels of IOPS and MBps performance,
integrating advanced technologies to maximize rotational velocities and data transfer rates,
while lowering seek times and latency. In addition, the Fibre Channel interface provides
robust functionality to process multiple I/O operations concurrently of varying sizes in both
directions at once.
The FATA slower drive mechanisms result in both lower IOPS and MBps performance
compared to Fibre Channel:
The FATA drive is not designed for fast access to data or handling large amounts of random
I/O. However, FATA drives are a good fit for many bandwidth applications because they can
provide comparable throughput for short periods of time.
Access frequency
In addition to random and sequential access patterns, another consideration is access
frequency and its relationship with secondary storage. Several secondary storage
implementations identified as ideal for FATA technology generate random data access, which
on the surface does not fit the FATA performance profile. But these implementations, such as
fixed content and reference data, will have sporadic access activity on large quantities of data
and will therefore primarily be measured by cost per gigabyte and not performance. Many
non-traditional IT environments, such as high-performance computing, rich media, and
energy, will significantly benefit from enterprise-class FATA solutions. These businesses are
looking for high throughput performance at the lowest cost per gigabyte, which is exactly what
FATA can deliver.
Backup application
The secondary storage implementation that fits the FATA performance profile exceptionally
well is backup, which generates sequential I/O as it streams data to the backup target, a
performance strength of FATA.
The backup of secondary data can be achieved more efficiently when the near-line storage
device acts as a caching device between Fibre Channel (FC) disks and tape, allowing the
primary disk to remain online longer. Advantages of this backup method is that it is faster and
consumes less server CPU than direct backup to tape.
Network
Near-line storage allows disk-to-disk backups to help achieve the following benefits:
Shorter backup time/higher application availability
Any IT department will tell you that its backup windows are either shrinking or already
nonexistent. As a result, IT personnel are always looking for ways to improve backup times
and minimize the amount of time a given application is affected by backup, either total
down time or time running in a degraded mode. By using disk as the backup target, the
backup runs and completes faster. After the data is safely stored on disk, the application is
free of the backup overhead. In addition, the data can then be moved to tape to provide
the long-term benefits of the traditional backup process.
Faster recovery time
In the past, tape was the only means of restoring data. This is a prolonged process,
because the appropriate tape has to be located, loaded into the tape drive, and then
sequentially read to locate and retrieve the desired data. Information has become
increasingly vital to a company’s success, and the lengthy restoration time from tape can
now be avoided. Backing up data to disk, as a disk image, enables significantly faster
restoration times, because data is stored online and can be located and retrieved
immediately.
Improved backup/restore reliability
Disk-to-disk backups create a new confidence in the ability to recover critical data by
eliminating the mechanical concerns associated with tape; one bad tape can cause large
restores to fail. Disk backups offer the same high level of RAID protection and redundancy
as the original data.
Easier backup/restore management
Storage management software functionality can be used to create volume-level copies, or
clones, of data as a source for restoration. Disk-to-disk backup packages, however,
provide more intelligence and file-level information that enables simplified administration
and faster restores.
Disk
Server Online Tape
Storage
Network
Disk
Reference Offline
Data Storage
Storage
Figure 18 Reference data storage scenario
Data retention
Recent government regulations have made it necessary to store, identify, and characterize
data. The majority of this data will be unchanging and accessed infrequently, if ever. As a
result, the highest possible performance is not a requirement. These implementations require
the largest amount of storage for the least cost in the least amount of space. The FATA cost
per gigabyte advantage over Fibre Channel and high capacity drives make it an attractive
solution.
Temporary workspace
FATA is a great fit for project-based applications that need short-term, affordable capacity.
Conclusion
We have discussed when FATA is a good choice depending on the nature of an application or
the type of storage required.
Important: IBM recommends that FATA drives be employed strictly with applications such
as those discussed in “The right FATA application” on page 60. Other types of applications,
and in particular transaction processing, must be avoided.
An important factor to keep in mind is that the FATA drives used in the DS8000 will protect
themselves by throttling IO based on the temperature registered by the internal sensors.
When throttled, the performance of the drives can drop by up to 50%, resulting in much higher
disk access times until the disk is able to return to its nominal temperature.
Important: Due to the lower duty cycle and the potential for I/O throttling, FATA based
volumes should not be mirrored with FC based volumes.
Also, to keep the FATA drives from robbing the enterprise disk traffic of DS8000 cache
resources, the modified writes to Non-Volatile Storage (NVS) for the FATA arrays are limited.
The user or storage administrator is responsible for targeting appropriate workloads to the
FATA drives.
Table 4-2 shows some typical applications and the most suitable DS8000 drive type.
Archiving and data retention Storage Capacity, high density Near-line FATA
Backup and recovery (disk Storage Capacity, high density Near-line FATA
to disk)
Database & data mining Mix of performance and On-line Fibre Channel
capacity
Temporary storage, spool, High performance and good On-line Fibre Channel
paging availability
An ESCON link consists of two fibers, one for each direction, connected at each end by an
ESCON connector to an ESCON port. Each ESCON adapter card supports two ESCON
ports or links, and each link supports 64 logical paths.
ESCON distances
For connections without repeaters, the ESCON distances are 2 km with 50 micron multimode
fiber, and 3 km with 62.5 micron multimode fiber. The DS8000 supports all models of the IBM
9032 ESCON directors that can be used to extend the cabling distances.
This site should be consulted regularly because it has the most up-to-date information on
server attachment support.
Each DS8000 Fibre Channel card offers four Fibre Channel ports (port speed of 2 or 4 Gbps
depending on the host adapter). The cable connector required to attach to this card is an LC
type. Each 2 Gbps port independently auto-negotiates to either 2 or 1 Gbps and the 4 Gbps
ports to 4 or 2 Gbps link speed. Each of the 4 ports on one DS8000 adapter can also
independently be either Fibre Channel protocol (FCP) or FICON, though the ports are initially
defined as switched point to point FCP. Selected ports will be configured to FICON
automatically based on the definition of a FICON host. Each port can be either FICON or
Fibre Channel protocol (FCP). The personality of the port is changeable via the DS Storage
Manager GUI. A port cannot be both FICON and FCP simultaneously, but it can be changed
as required.
The card itself is PCI-X 64 Bit 133 MHz. The card is driven by a new high function, high
performance ASIC. To ensure maximum data integrity, it supports metadata creation and
checking. Each Fibre Channel port supports a maximum of 509 host login IDs and 1280
paths. This allows for the creation of very large storage area networks (SANs). The design of
the card is depicted in Figure 4-19.
Processor
QDR
PPC 1 GHz
750GX
Fibre Channel
Protocol
Engine
http://www-03.ibm.com/servers/storage/disk/ds8000/interop.html
This document should be consulted regularly because it has the most up-to-date information
on server attachment support.
RPCs also communicate with each primary power supply and indirectly with each rack’s fan
sense cards and the disk enclosures in each frame.
There are two redundant PPSs in each frame of the DS8000. Each PPS is capable of
powering the frame by itself. The PPS creates 208V output power for the processor complex
and I/O enclosure power supplies. It also creates 5V and 12V DC power for the disk
enclosures. There may also be an optional booster module that will allow the PPSs to
temporarily run the disk enclosures off battery, if the extended power line disturbance feature
has been purchased (see Chapter 5, “RAS - Reliability, Availability, Serviceability” on
page 69, for a complete explanation as to why this feature may or may not be necessary for
your installation).
Each PPS has internal fans to supply cooling for that power supply.
The FC-AL DDMs are not protected from power loss unless the extended power line
disturbance feature has been purchased.
S-HMC
The S-HMC is the focal point for configuration, Copy Services management, and
maintenance activities. It is possible to order two management consoles to act as a redundant
pair. A typical configuration would be to have one internal and one external management
console. The internal S-HMC will contain a PCI modem for remote service.
Ethernet switches
In addition to the Fibre Channel switches installed in each disk enclosure, the DS8000 base
frame contains two 16-port Ethernet switches. Two switches are supplied to allow the creation
of a fully redundant management network. Each processor complex has multiple connections
to each switch. This is to allow each server to access each switch. This switch cannot be used
for any equipment not associated with the DS8000. The switches get power from the internal
power bus and thus do not require separate power outlets.
5.1 Naming
It is important to understand the naming conventions used to describe DS8000 components
and constructs in order to fully appreciate the discussion of RAS concepts.
Storage complex
This term describes a group of DS8000s managed by a single Management Console. A
storage complex may consist of just a single DS8000 storage unit.
Storage unit
A storage unit consists of a single DS8000 (including expansion frames). If your organization
has one DS8000, then you have a single storage complex that contains a single storage unit.
Processor Processor
complex 0 complex 1
Storage
server 0 facility server 1
image 1
LPARs
Figure 5-1 Single image mode
complex is used to form a storage image. If there are four servers, there are effectively two
separate storage subsystems existing inside one DS8300 model 9B2 storage unit.
LPARs
Processor Processor
complex 0 complex 1
Storage
Server 0 facility Server 1
image 1
Storage
Server 0 facility Server 1
image 2
LPARs
Figure 5-2 DS8300 Turbo Model 9B2 - dual image mode
In Figure 5-2 we have two storage facility images (SFIs). The upper server 0 and upper server
1 form SFI 1. The lower server 0 and lower server 1 form SFI 2. In each SFI, server 0 is the
darker color (green) and server 1 is the lighter color (yellow). SFI 1 and SFI 2 may share
common hardware (the processor complexes) but they are completely separate from an
operational point of view.
Note: You may think that the lower server 0 and lower server 1 should be called server 2
and server 3. While this may make sense from a numerical point of view (for example,
there are four servers so why not number them from 0 to 3), but each SFI is not aware of
the other’s existence. Each SFI must have a server 0 and a server 1, regardless of how
many SFIs or servers there are in a DS8000 storage unit.
For more information on DS8000 series storage system partitions see Chapter 3, “Storage
system LPARs (logical partitions)” on page 27.
Processor complex
A processor complex is one System p system unit. Two processor complexes form a
redundant pair such that if either processor complex fails, the servers on the remaining
processor complex can continue to run the storage image. In an ESS 800, we would have
referred to a processor complex as a cluster.
Fault avoidance
POWER5 systems are built to keep errors from ever happening. This quality-based design
includes such features as reduced power consumption and cooler operating temperatures for
increased reliability, enabled by the use of copper chip circuitry, SOI (silicon on insulator), and
dynamic clock-gating. It also uses mainframe-inspired components and technologies.
Any errors that are detected by the pervasive error checkers are captured into Fault Isolation
Registers (FIRs), which can be interrogated by the service processor (SP). The SP in the
System p5 has the capability to access system components using special-purpose service
processor ports or by access to the error registers.
The FIRs are important because they enable an error to be uniquely identified, thus enabling
the appropriate action to be taken. Appropriate actions might include such things as a bus
retry, ECC (error checking and correction), or system firmware recovery routines. Recovery
routines could include dynamic deallocation of potentially failing components.
Errors are logged into the system non-volatile random access memory (NVRAM) and the SP
event history log, along with a notification of the event to AIX for capture in the operating
system error log. Diagnostic Error Log Analysis (diagela) routines analyze the error log
entries and invoke a suitable action, such as issuing a warning message. If the error can be
recovered, or after suitable maintenance, the service processor resets the FIRs so that they
can accurately record any future errors.
The ability to correctly diagnose any pending or firm errors is a key requirement before any
dynamic or persistent component deallocation or any other reconfiguration can take place.
Permanent monitoring
The SP that is included in the System p5 provides a way to monitor the system even when the
main processor is inoperable. The next subsection offers a more detailed description of the
monitoring functions in the System p5.
Mutual surveillance
The SP can monitor the operation of the firmware during the boot process, and it can monitor
the operating system for loss of control. This enables the service processor to take
appropriate action when it detects that the firmware or the operating system has lost control.
Mutual surveillance also enables the operating system to monitor for service processor
activity and can request a service processor repair action if necessary.
Environmental monitoring
Environmental monitoring related to power, fans, and temperature is performed by the
System Power Control Network (SPCN). Environmental critical and non-critical conditions
generate Early Power-Off Warning (EPOW) events. Critical events (for example, a Class 5 AC
power loss) trigger appropriate signals from hardware to the affected components to prevent
any data loss without operating system or firmware involvement. Non-critical environmental
events are logged and reported using Event Scan. The operating system cannot program or
access the temperature threshold using the SP.
Temperature monitoring is also performed. If the ambient temperature goes above a preset
operating range, then the rotation speed of the cooling fans can be increased. Temperature
monitoring also warns the internal microcode of potential environment-related problems. An
orderly system shutdown will occur when the operating temperature exceeds a critical level.
Voltage monitoring provides warning and an orderly system shutdown when the voltage is out
of operational specification.
Self-healing
For a system to be self-healing, it must be able to recover from a failing component by first
detecting and isolating the failed component. It should then be able to take it offline, fix or
isolate it, and then reintroduce the fixed or replaced component into service without any
application disruption. Examples include:
Bit steering to redundant memory in the event of a failed memory module to keep the
server operational
Bit scattering, thus allowing for error correction and continued operation in the presence of
a complete chip failure (Chipkill recovery)
Single-bit error correction using ECC without reaching error thresholds for main, L2, and
L3 cache memory
L3 cache line deletes extended from 2 to 10 for additional self-healing
ECC extended to inter-chip connections on fabric and processor bus
Memory scrubbing to help prevent soft-error memory faults
Dynamic processor deallocation
The memory DIMMs also utilize memory scrubbing and thresholding to determine when
memory modules within each bank of memory should be used to replace ones that have
exceeded their threshold of error count (dynamic bit-steering). Memory scrubbing is the
process of reading the contents of the memory during idle time and checking and correcting
any single-bit errors that have accumulated by passing the data through the ECC logic. This
function is a hardware function on the memory controller chip and does not influence normal
system memory performance.
N+1 redundancy
The use of redundant parts, specifically the following ones, allows the System p5 to remain
operational with full resources:
Redundant spare memory bits in L1, L2, L3, and main memory
Redundant fans
Redundant power supplies
Fault masking
If corrections and retries succeed and do not exceed threshold limits, the system remains
operational with full resources and no client or IBM Service Representative intervention is
required.
Resource deallocation
If recoverable errors exceed threshold limits, resources can be deallocated with the system
remaining operational, allowing deferred maintenance at a convenient time.
Following a hardware error that has been flagged by the service processor, the subsequent
reboot of the server invokes extended diagnostics. If a processor or L3 cache has been
marked for deconfiguration by persistent processor deallocation, the boot process will attempt
to proceed to completion with the faulty device automatically deconfigured. Failing I/O
adapters will be deconfigured or bypassed during the boot process.
Concurrent Maintenance
Concurrent Maintenance provides replacement of the following parts while the processor
complex remains running:
Disk drives
Cooling fans
Power Subsystems
PCI-X adapter cards
memory, internal SCSI disks and other media bays (such as DVD-RAM), but also hardware
common between the two processor complexes. This hardware includes such things as the
I/O enclosures and the adapters installed within them.
A mechanism must exist to allow this sharing of resources in a seamless way. This
mechanism is called the hypervisor.
In addition, each I/O enclosure has N+1 power and cooling in the form of two power supplies
with integrated fans. The power supplies can be concurrently replaced and a single power
supply is capable of supplying DC power to an I/O drawer.
To create logical volumes on the DS8000, we work through the following constructs:
DDMs that are installed into pre-defined array sites.
These array sites are used to form RAID-5 or RAID-10 arrays.
These RAID arrays then become members of a rank.
Each rank then becomes a member of an extent pool. Each extent pool has an affinity to
either server 0 or server 1. Each extent pool is either open systems FB (fixed block) or
System z CKD (count key data).
Within each extent pool we create logical volumes, which for open systems are called
LUNs and for System z, 3390 volumes. LUN stands for logical unit number, which is used
for SCSI addressing. Each logical volume belongs to a logical subsystem (LSS).
For open systems the LSS membership is not that important (unless you are using Copy
Services), but for System z, the LSS is the logical control unit (LCU) which equates to a 3990
(a System z disk controller which the DS8000 emulates). What is important, is that LSSs that
have an even identifying number have an affinity with server 0, while LSSs that have an odd
identifying number have an affinity with server 1. When a host operating system issues a write
to a logical volume, the DS8000 host adapter directs that write to the server that owns the
LSS of which that logical volume is a member.
If the DS8000 is being used to operate a single storage image then the following examples
refer to two servers, one running on each processor complex. If a processor complex were to
fail then one server would fail. Likewise, if a server itself were to fail, then it would have the
same effect as the loss of the processor complex it runs on.
If, however, the DS8000 is divided into two storage images, then each processor complex will
be hosting two servers. In this case, a processor complex failure would result in the loss of
two servers. The effect on each server would be identical. The failover processes performed
by each storage image would proceed independently.
Data flow
When a write is issued to a volume, this write normally gets directed to the server that owns
this volume. The data flow is that the write is placed into the cache memory of the owning
server. The write data is also placed into the NVS memory of the alternate server.
NVS NVS
for odd for even
LSSs LSSs
Cache Cache
memory memory
for even for odd
LSSs LSSs
Server 0 Server 1
Figure 5-3 illustrates how the cache memory of server 0 is used for all logical volumes that
are members of the even LSSs. Likewise, the cache memory of server 1 supports all logical
volumes that are members of odd LSSs. But for every write that gets placed into cache,
another copy gets placed into the NVS memory located in the alternate server. Thus the
normal flow of data for a write is:
1. Data is written to cache memory in the owning server.
2. Data is written to NVS memory of the alternate server.
3. The write is reported to the attached host as having been completed.
4. The write is destaged from the cache memory to disk.
5. The write is discarded from the NVS memory of the alternate server.
Under normal operation, both DS8000 servers are actively processing I/O requests. This
section describes the failover and failback procedures that occur between the DS8000
servers when an abnormal condition has affected one of them.
Failover
In the example depicted in Figure 5-4 on page 78, server 0 has failed. The remaining server
has to take over all of its functions. The RAID arrays, because they are connected to both
servers, can be accessed from the device adapters used by server 1.
From a data integrity point of view, the real issue is the un-destaged or modified data that
belonged to server 1 (that was in the NVS of server 0). Since the DS8000 now has only one
copy of that data (which is currently residing in the cache memory of server 1), it will now take
the following steps:
1. It destages the contents of its NVS to the disk subsystem.
2. The NVS and cache of server 1 are divided in two, half for the odd LSSs and half for the
even LSSs.
3. Server 1 now begins processing the writes (and reads) for all the LSSs.
NVS NVS
NVS for for
for odd even odd
LSSs LSSs LSSs
Server 0 Server 1
Failover
Figure 5-4 Server 0 failing over its function to server 1
This entire process is known as a failover. After failover the DS8000 now operates as
depicted in Figure 5-4. Server 1 now owns all the LSSs, which means all reads and writes will
be serviced by server 1. The NVS inside server 1 is now used for both odd and even LSSs.
The entire failover process should be invisible to the attached hosts, apart from the possibility
of some temporary disk errors.
Failback
When the failed server has been repaired and restarted, the failback process is activated.
Server 1 starts using the NVS in server 0 again, and the ownership of the even LSSs is
transferred back to server 0. Normal operations with both controllers active then resumes.
Just like the failover process, the failback process is invisible to the attached hosts.
In general, recovery actions on the DS8000 do not impact I/O operation latency by more than
15 seconds. With certain limitations on configurations and advanced functions, this impact to
latency can be limited to 8 seconds. On logical volumes that are not configured with RAID-10
storage, certain RAID-related recoveries may cause latency impacts in excess of 15 seconds.
If you have real time response requirements in this area, contact IBM to determine the latest
information on how to manage your storage to meet your requirements,
The single purpose of the batteries is to preserve the NVS area of server memory in the event
of a complete loss of input power to the DS8000. If both power supplies in the base frame
were to stop receiving input power, the servers would be informed that they were now running
on batteries and immediately begin a shutdown procedure. Unless the power line disturbance
feature has been purchased, the BBUs are not used to keep the disks spinning. Even if they
do keep spinning, the design is to not move the data from NVS to the FC-AL disk arrays.
Instead, each processor complex has a number of internal SCSI disks which are available to
store the contents of NVS. When an on-battery condition related shutdown begins, the
following events occur:
1. All host adapter I/O is blocked.
2. Each server begins copying its NVS data to internal disk. For each server, two copies are
made of the NVS data in that server.
3. When the copy process is complete, each server shuts down.
4. When shutdown in each server is complete (or a timer expires), the DS8000 is powered
down.
An important point is that the servers will not come online until the batteries are sufficiently
charged to at least handle one outage (typically within a few minutes). In many cases,
sufficient charging will occur during the power on self test and storage image initialization.
However, if a complete discharge of the batteries has occurred, which may happen if multiple
power outages occur in a short period of time, then recharging may take up to two hours.
Because the contents of NVS are written to the internal SCSI disks of the DS8000 processor
complex and not held in battery protected NVS-RAM, the contents of NVS can be preserved
indefinitely. This means that unlike the DS6000 or the ESS 800, you are not held to a fixed
limit of time before power must be restored.
HBA
HP HP HP HP HP HP HP HP HP HP HP HP
Server
owning all RIO-G RIO-G I/O enclosure 2 RIO-G RIO-G
Server
owning all
even LSS odd LSS
logical logical
RIO-G RIO-G RIO-G RIO-G
volumes volumes
I/O enclosure 3
Slot 1 Slot 2 Slot 4 Slot 5
Fibre Fibre
channel channel ESCON ESCON
HP HP HP HP HP HP HP HP HP HP HP HP
It is always preferable that hosts that access the DS8000 have at least two connections to
separate host ports in separate host adapters on separate I/O enclosures, as in Figure 5-6.
HBA HBA
HP HP HP HP HP HP HP HP HP HP HP HP
Server
RIO-G
I/O enclosure 2 RIO-G RIO-G
Server
owning all RIO-G owning all
even LSS odd LSS
logical logical
RIO-G RIO-G RIO-G RIO-G
volumes volumes
I/O enclosure 3
Slot 1 Slot 2 Slot 4 Slot 5
Fibre Fibre
channel ESCON ESCON
channel
HP HP HP HP HP HP HP HP HP HP HP HP
In this example, the host is attached to different Fibre Channel host adapters in different I/O
enclosures. This is also important because during a microcode update, an I/O enclosure may
need to be taken offline. This configuration allows the host to survive a hardware failure on
any component on either path.
SAN/FICON/ESCON switches
Because a large number of hosts may be connected to the DS8000, each using multiple
paths, the number of host adapter ports that are available in the DS8000 may not be sufficient
to accommodate all the connections. The solution to this problem is the use of SAN switches
or directors to switch logical connections from multiple hosts. In a System z environment you
will need to select a SAN switch or director that also supports FICON. ESCON-attached hosts
may need an ESCON director.
A logic or power failure in a switch or director can interrupt communication between hosts and
the DS8000. We recommend that more than one switch or director be provided to ensure
continued availability. Ports from two different host adapters in two different I/O enclosures
should be configured to go through each of two directors. The complete failure of either
director leaves half the paths still operating.
Multi-pathing software
Each attached host operating system requires a mechanism to allow it to manage multiple
paths to the same device, and to preferably load balance these requests. Also, when a failure
occurs on one redundant path, then the attached host must have a mechanism to allow it to
detect that one path is gone and route all I/O requests for those logical devices to an
alternative path. Finally, it should be able to detect when the path has been restored so that
the I/O can again be load balanced. The mechanism that will be used varies by attached host
operating system and environment as detailed in the next two sections.
SDD provides availability through automatic I/O path failover. If a failure occurs in the data
path between the host and the DS8000, SDD automatically switches the I/O to another path.
SDD will also automatically set the failed path back online after a repair is made. SDD also
improves performance by sharing I/O operations to a common disk over multiple active paths
to distribute and balance the I/O workload. For the DS6000 and SVC, SDD also supports the
concept of preferred path.
SDD is not available for every supported operating system. Refer to the IBM System Storage
DS8000 Host Systems Attachment Guide, SC26-7917, and the interoperability Web site for
direction as to which multi-pathing software may be required. Some devices, such as the IBM
SAN Volume Controller (SVC), do not require any multi-pathing software because the internal
software in the device already supports multi-pathing. The interoperability Web site is:
http://www.ibm.com/servers/storage/disk/ds8000/interop.html
For more information on the SDD see Section 15.1.4, “Multipathing support — Subsystem
Device Driver (SDD)” on page 290.
These functions are part of the System z architecture and are managed by the channel
subsystem in the host and the DS8000.
A physical FICON/ESCON path is established when the DS8000 port sees light on the fiber
(for example, a cable is plugged in to a DS8000 host adapter, a processor or the DS8000 is
powered on, or a path is configured online by OS/390). At this time, logical paths are
established through the port between the host and some or all of the LCUs in the DS8000,
controlled by the HCD definition for that host. This happens for each physical path between a
System z CPU and the DS8000. There may be multiple system images in a CPU. Logical
paths are established for each system image. The DS8000 then knows which paths can be
used to communicate between each LCU and each host.
CUIR
Control Unit Initiated Reconfiguration (CUIR) prevents loss of access to volumes in System z
environments due to wrong path handling. This function automates channel path
management in System z environments, in support of selected DS8000 service actions.
Control Unit Initiated Reconfiguration is available for the DS8000 when operated in the z/OS
and z/VM environments. The CUIR function automates channel path vary on and vary off
actions to minimize manual operator intervention during selected DS8000 service actions.
CUIR allows the DS8000 to request that all attached system images set all paths required for
a particular service action to the offline state. System images with the appropriate level of
software support will respond to such requests by varying off the affected paths, and either
notifying the DS8000 subsystem that the paths are offline, or that it cannot take the paths
offline. CUIR reduces manual operator intervention and the possibility of human error during
maintenance actions, at the same time reducing the time required for the maintenance. This
is particularly useful in environments where there are many systems attached to a DS8000.
has a path to each switch, so it also can tolerate the loss of a single path. If both paths from
one DA fail, then it cannot access the switches; however, the other DA retains connection.
Server 1
device adapter
Storage enclosure
Midplane backplane
Figure 5-7 also shows the connection paths for expansion on the far left and far right. The
paths from the switches travel to the switches in the next disk enclosure. Because expansion
is done in this linear fashion, the addition of more enclosures is completely non-disruptive.
RAID-5 theory
The DS8000 series supports RAID-5 arrays. RAID-5 is a method of spreading volume data
plus parity data across multiple disk drives. RAID-5 provides faster performance by striping
data across a defined set of DDMs. Data protection is provided by the generation of parity
information for every stripe of data. If an array member fails, then its contents can be
regenerated by using the parity data.
Drive failure
When a disk drive module fails in a RAID-5 array, the device adapter starts an operation to
reconstruct the data that was on the failed drive onto one of the spare drives. The spare that
is used will be chosen based on a smart algorithm that looks at the location of the spares and
the size and location of the failed DDM. The rebuild is performed by reading the
corresponding data and parity in each stripe from the remaining drives in the array,
performing an exclusive-OR operation to recreate the data, then writing this data to the spare
drive.
While this data reconstruction is going on, the device adapter can still service read and write
requests to the array from the hosts. There may be some degradation in performance while
the sparing operation is in progress because some DA and switched network resources are
being used to do the reconstruction. Due to the switch-based architecture, this effect will be
minimal. Additionally, any read requests for data on the failed drive requires data to be read
from the other drives in the array and then the DA performs an operation to reconstruct the
data.
Performance of the RAID-5 array returns to normal when the data reconstruction onto the
spare device completes. The time taken for sparing can vary, depending on the size of the
failed DDM and the workload on the array, the switched network, and the DA. The use of
arrays across loops (AAL) both speeds up rebuild time and decreases the impact of a rebuild.
RAID-10 theory
RAID-10 provides high availability by combining features of RAID-0 and RAID-1. RAID-0
optimizes performance by striping volume data across multiple disk drives at a time. RAID-1
provides disk mirroring, which duplicates data between two disk drives. By combining the
features of RAID-0 and RAID-1, RAID-10 provides a second optimization for fault tolerance.
Data is striped across half of the disk drives in the RAID-1 array. The same data is also
striped across the other half of the array, creating a mirror. Access to data is preserved if one
disk in each mirrored pair remains available. RAID-10 offers faster data reads and writes than
RAID-5 because it does not need to manage parity. However, with half of the DDMs in the
group used for data and the other half to mirror that data, RAID-10 disk groups have less
capacity than RAID-5 disk groups.
Drive failure
When a disk drive module (DDM) fails in a RAID-10 array, the controller starts an operation to
reconstruct the data from the failed drive onto one of the hot spare drives. The spare that is
used will be chosen based on a smart algorithm that looks at the location of the spares and
the size and location of the failed DDM. Remember a RAID-10 array is effectively a RAID-0
array that is mirrored. Thus when a drive fails in one of the RAID-0 arrays, we can rebuild the
failed drive by reading the data from the equivalent drive in the other RAID-0 array.
While this data reconstruction is going on, the DA can still service read and write requests to
the array from the hosts. There may be some degradation in performance while the sparing
operation is in progress because some DA and switched network resources are being used to
do the reconstruction. Due to the switch-based architecture of the DS8000, this effect will be
minimal. Read requests for data on the failed drive should not be affected because they can
all be directed to the good RAID-1 array.
Write operations will not be affected. Performance of the RAID-10 array returns to normal
when the data reconstruction onto the spare device completes. The time taken for sparing
can vary, depending on the size of the failed DDM and the workload on the array and the DA.
On the ESS 800 the spare creation policy was to have four DDMs on each SSA loop for each
DDM type. This meant that on a specific SSA loop it was possible to have 12 spare DDMs if
you chose to populate a loop with three different DDM sizes. With the DS8000 the intention is
to not do this. A minimum of one spare is created for each array site defined until the following
conditions are met:
A minimum of 4 spares per DA pair
A minimum of 4 spares of the largest capacity array site on the DA pair
A minimum of 2 spares of capacity and RPM greater than or equal to the fastest array site
of any given capacity on the DA pair
Floating spares
The DS8000 implements a smart floating technique for spare DDMs. On an ESS 800, the
spare floats. This means that when a DDM fails and the data it contained is rebuilt onto a
spare, then when the disk is replaced, the replacement disk becomes the spare. The data is
not migrated to another DDM, such as the DDM in the original position the failed DDM
occupied. So in other words, on an ESS 800 there is no post repair processing.
The DS8000 microcode may choose to allow the hot spare to remain where it has been
moved, but it may instead choose to migrate the spare to a more optimum position. This will
be done to better balance the spares across the DA pairs, the loops, and the enclosures. It
may be preferable that a DDM that is currently in use as an array member be converted to a
spare. In this case the data on that DDM will be migrated in the background onto an existing
spare. This process does not fail the disk that is being migrated, though it does reduce the
number of available spares in the DS8000 until the migration process is complete.
A smart process will be used to ensure that the larger or higher RPM DDMs always act as
spares. This is preferable because if we were to rebuild the contents of a 146 GB DDM onto a
300 GB DDM, then approximately half of the 300 GB DDM will be wasted since that space is
not needed. The problem here is that the failed 146 GB DDM will be replaced with a new
146 GB DDM. So the DS8000 microcode will most likely migrate the data back onto the
recently replaced 146 GB DDM. When this process completes, the 146 GB DDM will rejoin
the array and the 300 GB DDM will become the spare again. Another example would be if we
fail a 73 GB 15k RPM DDM onto a 146 GB 10k RPM DDM. This means that the data has now
moved to a slower DDM, but the replacement DDM will be the same as the failed DDM. This
means the array will have a mix of RPMs. This is not desirable. Again, a smart migrate of the
data will be performed once suitable spares have become available.
208V is produced to be supplied to each I/O enclosure and each processor complex. This
voltage is placed by each supply onto two redundant power buses.
12V and 5V is produced to be supplied to the disk enclosures.
If either PPS fails, the other can continue to supply all required voltage to all power buses in
that frame. The PPS can be replaced concurrently.
Important: It should be noted that if you install the DS8000 such that both primary power
supplies are attached to the same circuit breaker or the same switch board, then the
DS8000 will not be well protected from external power failures. This is a very common
cause of unplanned outages.
If building power is not considered reliable then the addition of the extended power line
disturbance feature should be considered. This feature adds two separate pieces of hardware
to the DS8000:
1. For each primary power supply in each frame of the DS8000, a booster module is added
that converts 208V battery power into 12V and 5V. This is to supply the DDMs with power
directly from the batteries. The PPSs do not normally receive power from the BBUs.
2. Batteries will be added to expansion racks that did not already have them. Base racks and
expansion racks with I/O enclosures get batteries by default. Expansion racks that do not
have I/O enclosures normally do not get batteries.
With the addition of this hardware, the DS8000 will be able to run for up to 50 seconds on
battery power, before the servers begin to copy NVS to SCSI disk and then shutdown. This
would allow for a 50 second interruption to building power with no outage to the DS8000.
Apart from these two contingencies (which are highly unlikely), the EPO switch should never
be used. The reason for this is that the DS8000 NVS storage area is not directly protected by
batteries. If building power is lost, the DS8000 can use its internal batteries to destage the
data from NVS memory to a variably sized disk area to preserve that data until power is
restored. However, the EPO switch does not allow this destage process to happen and all
NVS data is lost. This will most likely result in data loss.
If you need to power the DS8000 off for building maintenance, or to relocate it, you should
always use the S-HMC to achieve this.
For detailed discussion on microcode updates refer to Chapter 18, “Licensed machine code”
on page 375.
S-HMC
The S-HMC is used to perform configuration, management, and maintenance activities on the
DS8000. It can be ordered to be located either physically inside the base frame or external for
mounting in a customer-supplied rack.
If the S-HMC is not operational then it is not possible to perform maintenance, power the
DS8000 up or down, or perform Copy Services tasks such as the establishment of
FlashCopies. It is thus recommended to order two management consoles to act as a
redundant pair. Alternatively, if TotalStorage Productivity Center - Replication Manager (TPC
RM) is used, copy services tasks can be managed by that tool if the S-HMC is unavailable.
Ethernet switches
Each DS8000 base frame contains two 16-port Ethernet switches. Two switches are supplied
to allow the creation of a fully redundant management network. Each server in the DS8000
has a connection to each switch. Each S-HMC also has a connection to each switch. This
means that should a single Ethernet switch fail, all traffic can successfully travel from either
S-HMC to any server in the storage unit using the alternate switch.
IBM Service personnel located outside of the client facility log in to the DS HMC to provide
service and support.
The remote support as well as the call home options are described in detail in Section 20.1,
“Call Home and remote support” on page 392.
A storage unit rack with this optional seismic kit includes cross-braces on the front and rear of
the rack which prevent the rack from twisting. Hardware at the bottom of the rack secures it to
the floor. Depending on the flooring in your environment, specifically non-raised floors,
installation of required floor mounting hardware may be disruptive.
The Earthquake Resistance Kit is an optional feature for the DS8000 series.
One important feature of the DS8000 is the virtualization of a whole storage subsystem. For
example, if a service provider runs workloads for different banks, it would be best to
completely separate the workloads. This could be done on the host side with IBM’s LPAR
technology; the same technology is now also available for a storage system, the IBM DS8000.
For this chapter, the definition of virtualization is the abstraction process from the physical
disk drives to a logical volume that the hosts and servers see as if it were a physical disk.
Like in non-LPAR mode, where there are two SMPs running an AIX kernel and forming a
storage complex with two servers, server0 and server1, a SFI is a storage complex of its own.
Since it does not own the physical hardware (the storage unit), you can think of it as a virtual
storage system. Each SFI has a server 0 and a server 1 is totally separate from the other SFI
by the LPAR hypervisor. Disk drives and arrays are owned by one or the other SFI, they
cannot be shared. Figure 6-1 illustrates the storage LPAR concept. In the following section,
server 0 or server 1 could also mean server 0 or server 1 of a SFI running in an LPAR.
Logical
view: Storage Facility image 1 Storage Facility image 2
virtual
LIC RIO-G LIC LIC RIO-G LIC
Storage Memory
I/O I/O
Memory Memory
I/O I/O
Memory
takes part of
view: RIO-G
physical I/O I/O
storage Memory Memory
unit
Processor Processor
The DS8000 is populated with switched FC-AL disk drives that are mounted in disk
enclosures. You order disk drives in groups of 16 drives of the same capacity and RPM. The
disk drives can be accessed by a pair of device adapters. Each device adapter has four paths
to the disk drives. The four paths provide two FC-AL device interfaces, each with two paths,
such that either path can be used to communicate with any disk drive on that device interface
(in other words, the paths are redundant). One device interface from each device adapter is
connected to a set of FC-AL devices such that either device adapter has access to any disk
drive through two independent switched fabrics (in other words, the device adapters and
switches are redundant).
Each device adapter has four ports and since device adapters operate in pairs, there are
eight ports or paths to the disk drives. All eight paths can operate concurrently and could
access all disk drives on the attached fabric. In normal operation, however, disk drives are
typically accessed by one device adapter. Which device adapter owns the disk is defined
during the logical configuration process. This avoids any contention between the two device
adapters for access to the disks.
I/O Enclosure
I/O Enclosure
RIO-G HA HA DA HA HA DA
HA HA DA HA HA DA
Storage enclosure pair
Server0
Server1
Switches
Compare this with the ESS design, where there was a real loop and having an 8-pack close to
a device adapter was an advantage. This is no longer relevant for the DS8000. Because of
the switching design, each drive is in close reach of the device adapter, apart from a few more
hops through the Fibre Channel switches for some drives. So, it is not really a loop, but a
switched FC-AL loop with the FC-AL addressing schema: Arbitrated Loop Physical
Addressing (AL-PA).
Array
Site
Switch
Loop 1 Loop 2
Figure 6-3 Array site
As you can see from Figure 6-3, array sites span loops. Four DDMs are taken from loop 1 and
another four DDMs from loop 2. Array sites are the building blocks used to define arrays.
6.3.2 Arrays
An array is created from one array site. Forming an array means defining it as a specific
RAID type. The supported RAID types are RAID-5 and RAID-10 (see 5.6.2, “RAID-5
overview” on page 83 and 5.6.3, “RAID-10 overview” on page 84). For each array site you can
select a RAID type. The process of selecting the RAID type for an array is also called defining
an array.
Note: In the DS8000 implementation, one array is defined using one array site.
According to the DS8000 sparing algorithm, from zero to two spares may be taken from the
array site. This is discussed further in Section 5.6.4, “Spare creation” on page 85.
Figure 6-4 shows the creation of a RAID-5 array with one spare, also called a 6+P+S array
(capacity of 6 DDMs for data, capacity of one DDM for parity, and a spare drive). According to
the RAID-5 rules, parity is distributed across all seven drives in this example.
On the right-hand side in Figure 6-4 the terms D1, D2, D3, and so on, stand for the set of data
contained on one disk within a stripe on the array. If, for example, 1 GB of data is written, it is
distributed across all the disks of the array.
Array
Site
D1 D7 D13 ...
D2 D8 D14 ...
D3 D9 D15 ...
D6 P D17 ...
So, an array is formed using one array site, and while the array could be accessed by each
adapter of the device adapter pair, it is managed by one device adapter. Which adapter and
which server manages this array is be defined later in the configuration path.
6.3.3 Ranks
In the DS8000 virtualization hierarchy there is another logical construct, a rank. When
defining a new rank, its name is chosen by the DS Storage Manager, for example, R1, R2, or
R3, and so on. You have to add an array to a rank.
Note: In the DS8000 implementation, a rank is built using just one array.
The available space on each rank will be divided into extents. The extents are the building
blocks of the logical volumes. An extent is striped across all disks of an array as shown in
Figure 6-5 on page 96 and indicated by the small squares in Figure 6-6 on page 97.
People who work in the System z environment typically do not deal with gigabytes, instead
they think of storage in metrics of the original 3390 volume sizes. A 3390 Model 3 is three
times the size of a Model 1, and a Model 1 has 1113 cylinders which is about 0.94 GB. The
extent size of a CKD rank therefore was chosen to be one 3390 Model 1 or 1113 cylinders.
One extent is the minimum physical allocation unit when a LUN or CKD volume is created, as
we discuss later. It is still possible to define a CKD volume with a capacity that is an integral
multiple of one cylinder or a fixed block LUN with a capacity that is an integral multiple of 128
logical blocks (64K bytes). However, if the defined capacity is not an integral multiple of the
capacity of one extent, the unused capacity in the last extent is wasted. For instance, you
could define a 1 cylinder CKD volume, but 1113 cylinders (1 extent) is allocated and 1112
cylinders would be wasted.
Figure 6-5 shows an example of an array that is formatted for FB data with 1 GB extents (the
squares in the rank just indicate that the extent is composed of several blocks from different
DDMs).
Array
D4 D10 D16 ...
Data
D5 D11 P ...
Data
Parity D6 P D17 ...
Creation of
a Rank
....
FB Rank
....
....
of 1GB
....
.... extents
....
....
One or more ranks with the same extent type can be assigned to an extent pool. One rank can
be assigned to only one extent pool. There can be as many extent pools as there are ranks.
The DS Storage Manager GUI guides the user to use the same RAID types in an extent pool.
As such, when an extent pool is defined, it must be assigned with the following attributes:
– Server affinity
– Extent type
– RAID type
The minimum number of extent pools is one; however, normally it should be at least two with
one assigned to server 0 and the other to server 1 so that both servers are active. In an
environment where FB and CKD are to go onto the DS8000 storage server, four extent pools
would provide one FB pool for each server, and one CKD pool for each server, to balance the
capacity between the two servers. Figure 6-6 is an example of a mixed environment with CKD
and FB extent pools. Additional extent pools may also be desirable to segregate ranks with
different DDM types. Extent pools are expanded by adding more ranks to the pool. Ranks are
organized in two rank groups; rank group 0 is controlled by server 0 and rank group 1 is
controlled by server 1.
Important: Capacity should be balanced between the two servers for best performance.
Server1
1GB 1GB 1GB 1GB
FB FB FB FB
Extent Pool FBtest
On a DS8000 up to 65280 (we use the abbreviation 64K in this discussion, even though it is
actually 65536 - 256, which is not quite 64K in binary) volumes can be created (64K CKD, or
64K FB volumes, or a mix of both types, but the sum cannot exceed 64K).
LUNs can be allocated in binary GB (230 bytes), decimal GB (109 byes), or 512 or 520 byte
blocks. However, the physical capacity that is allocated for a LUN is always a multiple of 1 GB,
so it is a good idea to have LUN sizes that are a multiple of a gigabyte. If you define a LUN
with a LUN size that is not a multiple of 1 GB, for example, 25.5 GB, the LUN size is 25.5 GB,
but 26 GB are physically allocated and 0.5 GB of the physical storage is unusable.
CKD volumes
A System z CKD volume is composed of one or more extents from one CKD extent pool. CKD
extents are of the size of 3390 Model 1, which has 1113 cylinders. However, when you define
a System z CKD volume, you do not specify the number of 3390 Model 1 extents but the
number of cylinders you want for the volume.
You can define CKD volumes with up to 65520 cylinders, which is about 55.6 GB.
If the number of cylinders specified is not an integral multiple of 1113 cylinders, then some
space in the last allocated extent is wasted. For example, if you define 1114 or 3340
cylinders, 1112 cylinders are wasted. For maximum storage efficiency, you should consider
allocating volumes that are exact multiples of 1113 cylinders. In fact, integral multiples of 3339
cylinders should be consider for future compatibility.
If you want to use the maximum number of cylinders (65520), you should consider that this is
not a multiple of 1113. You could go with 65520 cylinders and waste 147 cylinders for each
volume (the difference to the next multiple of 1113) or you might be better off with a volume
size of 64554 cylinders which is a multiple of 1113 (factor of 58), or even better, with 63441
cylinders which is a multiple of 3339, a model 3 size.
A CKD volume cannot span multiple extent pools, but a volume can have extents from
different ranks in the same extent pool. Figure 6-7 shows how a logical volume is allocated
with a CKD volume as an example. The allocation process for FB volumes is very similar and
is shown in Figure 6-8.
Rank-b 1 GB 1 GB
used
free
used
free Allocate a 3 GB LUN
Rank-a 1 GB 1 GB 1 GB 1 GB
used
3 GB LUN 2.9 GB LUN
Rank-b
created
1 GB 1 GB
used used
used used
100 MB unused
System i LUNs
System i LUNs are also composed of fixed block 1 GB extents. There are, however, some
special aspects with System i LUNs. LUNs created on a DS8000 are always RAID protected.
LUNs are based on RAID-5 or RAID-10 arrays. However, you might want to deceive OS/400
and tell it that the LUN is not RAID protected. This causes OS/400 to do its own mirroring.
System i LUNs can have the attribute unprotected, in which case the DS8000 will lie to an
System i host and tell it that the LUN is not RAID protected.
OS/400 only supports certain fixed volume sizes, for example model sizes of 8.5 GB,
17™.5 GB, and 35.1 GB. These sizes are not multiples of 1 GB and hence, depending on the
model chosen, some space is wasted. System i LUNs expose a 520 byte block to the host.
The operating system uses 8 of these bytes so the usable space is still 512 bytes like other
SCSI LUNs. The capacities quoted for the System i LUNs are in terms of the 512 byte block
capacity and are expressed in GB (109 ). These capacities should be converted to GB (230 )
when considering effective utilization of extents that are 1 GB (230 ). For more information on
this topic see Appendix 17, “System i considerations” on page 343.
allocation of logical volumes across ranks to improve performance, except for the case in
which the logical volume needed is larger than the total capacity of the single rank.
This construction method of using fixed extents to form a logical volume in the DS8000 allows
flexibility in the management of the logical volumes. We can now delete LUNs and reuse the
extents of those LUNs to create other LUNs, maybe of different sizes. One logical volume can
be removed without affecting the other logical volumes defined on the same extent pool.
Compared to the ESS, where it was not possible to delete a LUN unless the whole array was
reformatted, this DS8000 implementation gives you much more flexibility and allows for on
demand changes according to your needs.
Since the extents are cleaned after you have deleted a LUN or CKD volume, it may take some
time until these extents are available for reallocation. The reformatting of the extents is a
background process.
On an ESS there was a fixed association between logical subsystems (and their associated
logical volumes) and device adapters (and their associated ranks). The association of an
8-pack to a device adapter determined what LSS numbers could be chosen for a volume. On
an ESS up to 16 LSSs could be defined depending on the physical configuration of device
adapters and arrays.
On the DS8000, there is no fixed binding between any rank and any logical subsystem. The
capacity of one or more ranks can be aggregated into an extent pool and logical volumes
configured in that extent pool are not bound to any specific rank. Different logical volumes on
the same logical subsystem can be configured in different extent pools. As such, the available
capacity of the storage facility can be flexibly allocated across the set of defined logical
subsystems and logical volumes.
This predetermined association between array and LSS is gone on the DS8000. Also the
number of LSSs has changed. You can now define up to 255 LSSs for the DS8000. You can
even have more LSSs than arrays.
For each LUN or CKD volume you can now choose an LSS. You can put up to 256 volumes
into one LSS. There is, however, one restriction. We already have seen that volumes are
formed from a bunch of extents from an extent pool. Extent pools, however, belong to one
server, server 0 or server 1, respectively. LSSs also have an affinity to the servers. All even
numbered LSSs (X’00’, X’02’, X’04’, up to X’FE’) belong to server 0 and all odd numbered
LSSs (X’01’, X’03’, X’05’, up to X’FD’) belong to server 1. LSS X’FF’ is reserved.
System z users are familiar with a logical control unit (LCU). System z operating systems
configure LCUs to create device addresses. There is a one to one relationship between an
LCU and a CKD LSS (LSS X'ab' maps to LCU X'ab'). Logical volumes have a logical volume
number X'abcd' where X'ab' identifies the LSS and X'cd' is one of the 256 logical volumes on
the LSS. This logical volume number is assigned to a logical volume when a logical volume is
created and determines the LSS that it is associated with. The 256 possible logical volumes
associated with an LSS are mapped to the 256 possible device addresses on an LCU (logical
volume X'abcd' maps to device address X'cd' on LCU X'ab'). When creating CKD logical
volumes and assigning their logical volume numbers, users should consider whether Parallel
Access Volumes (PAV) are required on the LCU and reserve some of the addresses on the
LCU for alias addresses. For more information on PAV see 16.3, “z/OS considerations” on
page 335.
For open systems, LSSs do not play an important role except in determining which server the
LUN is managed by (and which extent pools it must be allocated in) and in certain aspects
related to Metro Mirror, Global Mirror, or any of the other remote copy implementations.
Some management actions in Metro Mirror, Global Mirror, or Global Copy operate at the LSS
level. For example the freezing of pairs to preserve data consistency across all pairs, in case
you have a problem with one of the pairs, is done at the LSS level. With the option now to put
all or most of the volumes of a certain application in just one LSS, this makes the
management of remote copy operations easier; see Figure 6-9.
LSS X'17'
DB2
LSS X'18'
DB2-test
Of course you could have put all volumes for one application in one LSS on an ESS, too, but
then all volumes of that application would also be in one or a few arrays, and from a
performance standpoint this was not desirable. Now on the DS8000 you can group your
volumes in one or a few LSSs but still have the volumes in many arrays or ranks.
Fixed block LSSs are created automatically when the first fixed block logical volume on the
LSS is created and deleted automatically when the last fixed block logical volume on the LSS
is deleted. CKD LSSs require user parameters to be specified and must be created before the
first CKD logical volume can be created on the LSS; they must be deleted manually after the
last CKD logical volume on the LSS is deleted.
Address groups
Address groups are created automatically when the first LSS associated with the address
group is created, and deleted automatically when the last LSS in the address group is
deleted.
LSSs are either CKD LSSs or FB LSSs. All devices in an LSS must be either CKD or FB. This
restriction goes even further. LSSs are grouped into address groups of 16 LSSs. LSSs are
numbered X'ab', where a is the address group and b denotes an LSS within the address
group. So, for example X'10' to X'1F' are LSSs in address group 1.
All LSSs within one address group have to be of the same type, CKD or FB. The first LSS
defined in an address group fixes the type of that address group.
System z users who still want to use ESCON to attach hosts to the DS8000 should be aware
that ESCON supports only the 16 LSSs of address group 0 (LSS X'00' to X'0F'). Therefore
this address group should be reserved for ESCON-attached CKD devices, in this case, and
not used as FB LSSs.
Server1
X'1E01'
X'1D00'
LSS X'1E'
Extent Pool FB-1 LSS X'1F' Extent Pool FB-2
Rank-c Rank-y
The LUN identifications X'gabb' are composed of the address group X'g', and the LSS
number within the address group X'a’, and the position of the LUN within the LSS X'bb'. For
example, LUN X'2101' denotes the second (X'01') LUN in LSS X'21' of address group 2.
Host attachment
HBAs are identified to the DS8000 in a host attachment construct that specifies the HBAs’
World Wide Port Names (WWPNs). A set of host ports can be associated through a port
group attribute that allows a set of HBAs to be managed collectively. This port group is
referred to as host attachment within the GUI.
A given host attachment can be associated with only one volume group. Each host
attachment can be associated with a volume group to define which LUNs that HBA is allowed
to access. Multiple host attachments can share the same volume group. The host attachment
may also specify a port mask that controls which DS8000 I/O ports the HBA is allowed to log
in to. Whichever ports the HBA logs in on, it sees the same volume group that is defined in the
host attachment associated with this HBA.
Volume group
A volume group is a named construct that defines a set of logical volumes. When used in
conjunction with CKD hosts, there is a default volume group that contains all CKD volumes
and any CKD host that logs into a FICON I/O port has access to the volumes in this volume
group. CKD logical volumes are automatically added to this volume group when they are
created and automatically removed from this volume group when they are deleted.
When used in conjunction with Open Systems hosts, a host attachment object that identifies
the HBA is linked to a specific volume group. The user must define the volume group by
indicating which fixed block logical volumes are to be placed in the volume group. Logical
volumes may be added to or removed from any volume group dynamically.
There are two types of volume groups used with Open Systems hosts and the type
determines how the logical volume number is converted to a host addressable LUN_ID on the
Fibre Channel SCSI interface. A map volume group type is used in conjunction with FC SCSI
host types that poll for LUNs by walking the address range on the SCSI interface. This type of
volume group can map any FB logical volume numbers to 256 LUN_IDs that have zeroes in
the last six bytes and the first two bytes in the range of X'0000' to X'00FF'.
A mask volume group type is used in conjunction with FC SCSI host types that use the Report
LUNs command to determine the LUN_IDs that are accessible. This type of volume group
can allow any and all FB logical volume numbers to be accessed by the host where the mask
is a bitmap that specifies which LUNs are accessible. For this volume group type, the logical
volume number X'abcd' is mapped to LUN_ID X'40ab40cd00000000'. The volume group type
also controls whether 512 byte block LUNs or 520 byte block LUNs can be configured in the
volume group.
When associating a host attachment with a volume group, the host attachment contains
attributes that define the logical block size and the Address Discovery Method (LUN Polling or
Report LUNs) that are used by the host HBA. These attributes must be consistent with the
volume group type of the volume group that is assigned to the host attachment so that HBAs
that share a volume group have a consistent interpretation of the volume group definition and
have access to a consistent set of logical volume types. The GUI typically sets these values
appropriately for the HBA based on the user specification of a host type. The user must
consider what volume group type to create when setting up a volume group for a particular
HBA.
FB logical volumes may be defined in one or more volume groups. This allows a LUN to be
shared by host HBAs configured to different volume groups. An FB logical volume is
automatically removed from all volume groups when it is deleted.
WWPN-7
Host att: Prog
WWPN-8 Volume group: docs
Figure 6-11 shows the relationships between host attachments and volume groups. Host
AIXprod1 has two HBAs, which are grouped together in one host attachment, and both are
granted access to volume group DB2-1. Most of the volumes in volume group DB2-1 are also
in volume group DB2-2, accessed by server AIXprod2. In our example, there is, however, one
volume in each group that is not shared. The server in the lower left part has four HBAs and
they are divided into two distinct host attachments. One can access some volumes shared
with AIXprod1 and AIXprod2. The other HBAs have access to a volume group called “docs.”
Next we created logical volumes within the extent pools, assigning them a logical volume
number that determined which logical subsystem they would be associated with and which
server would manage them. Then the LUNs could be assigned to one or more volume
groups. Finally the host HBAs were configured into a host attachment that is associated with
a given volume group.
This new virtualization concept provides for much more flexibility. Logical volumes can
dynamically be created and deleted. They can be grouped logically to simplify storage
management. Large LUNs and CKD volumes reduce the total number of volumes and this
also contributes to a reduction of the management effort.
Data
1 GB FB
1 GB FB
1 GB FB
Data
1 GB FB
1 GB FB
1 GB FB
Data
Data
Server0
Data
1 GB FB
1 GB FB
1 GB FB
Data
Parity
Spare
X'2x' FB
4096
addresses
LSS X'27'
X'3x' CKD
4096
addresses
Loop 2
Loop 1
LSS 00 LSS 01
DA pair 1
Server1
DA pair 2
As explained in the previous chapters, there are several options on how to create logical
volumes. You can select an extent pool that is owned by one server. There could be just one
extent pool per server or you could have several. The ranks of extent pools could come from
arrays on different device adapter pairs and different loops or from the same loop. Figure 6-13
shows an optimal distribution of eight logical volumes within a DS8000. Of course you could
have more extent pools and ranks, but when you want to distribute your data for optimal
performance, you should make sure that you spread it across the two servers, across different
device adapter pairs, across the loops, and across several ranks.
If you use some kind of a logical volume manager (like LVM on AIX) on your host, you can
create a host logical volume from several DS8000 logical volumes (LUNs). You can select
LUNs from different DS8000 servers, device adapter pairs, and loops as shown in
Figure 6-13. By striping your host logical volume across the LUNs, you will get the best
performance for this LVM volume.
These functions make the DS8000 series a key component for disaster recovery solutions,
data migration activities, as well as for data duplication and backup solutions.
The information discussed in this chapter is covered in greater extent and detail in the
following redbooks:
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787
The Copy Services functions run on the DS8000 storage unit and support open systems and
System z environments. These functions are supported also in the DS6000 series and on the
previous generation of storage systems —the IBM TotalStorage Enterprise Storage Server
(ESS) models.
Many design characteristics of the DS8000 and its data copy and mirror capabilities features
contribute to the protection of your data, 24 hours a day and seven days a week.
We explain these interfaces in 7.4, “Interfaces for Copy Services” on page 124.
7.2 FlashCopy
FlashCopy is designed to provide a point-in-time copy capability for logical volumes with
minimal interruption to applications, and makes it possible to access both the source and
target copies immediately.
In this section we discuss FlashCopy basic characteristics and options. For a more detailed
and extensive discussion on these topics, you can refer to the redbooks: IBM System Storage
DS8000 Series: Copy Services in Open Environments, SG24-6788, and IBM System Storage
DS8000 Series: Copy Services with System z servers, SG24-6787.
Note: In this chapter, track means a piece of data in the DS8000; the DS8000 uses the
logical tracks to manage the Copy Services functions.
Source Target
FlashCopy command issued
If you access the source or the target volumes during the background copy, FlashCopy
manages these I/O requests as follows:
The background copy may have a slight impact to your application because the physical copy
needs some storage resources, but the impact is minimal because the host I/O is prior to the
background copy. And if you want, you can issue FlashCopy with the no background copy
option.
To use FlashCopy you must have the corresponding licensed function indicator feature in the
DS8000, and you must acquire the corresponding DS8000 function authorization with the
adequate feature number —license in terms of physical capacity. For details on features and
functions requirements, see Section 11.1, “DS8000 licensed functions” on page 198.
Note: for a detailed explanation of the features involved and the considerations you must
have when ordering FlashCopy we recommend you refer to the announcement letters:
IBM System Storage DS8000 — Function Authorization (IBM 239x or 2244)
IBM System Storage DS8000 Series (IBM 242x or 2107)
In many cases, at most 10 to 20 percent of the entire data is changed in a day. In such a
situation, if you use this function for daily backup, you can save the time for the physical copy
of FlashCopy.
Initial FlashCopy
Incremental FlashCopy relationship established
with change recording and
Write
Write Source Target persistent copy options
Read
Read Control bitmap for each
volume created
Incremental FlashCopy
started
Tracks changed on the
target are overwritten by
the corresponding tracks
from the source
Tracks changed on the
source are copied to the
target
Possible reverse operation,
the target updates the
source
Figure 7-2 Incremental FlashCopy
You can also issue incremental FlashCopy from the target volume to the source volumes with
the reverse restore option. The reverse restore operation cannot be done unless the
background copy in the original direction has finished.
Dataset
Dataset
Note: If a FlashCopy source volume has more than one target, that source volume can be
involved only in a single incremental FlashCopy relationship.
Source Target
Volume Volume
1’
1 2’
12’
In order to create this consistent copy, the user would issue a set of establish FlashCopy
commands with a freeze option, which will hold off host I/O to the source volumes. In other
words, Consistency Group FlashCopy provides the capability to temporarily queue (at the
host I/O level, not the application level) subsequent write operations to the source volumes
that are part of the Consistency Group. During the temporary queueing, Establish FlashCopy
is completed. The temporary queueing continues until this condition is reset by the
Consistency Group Created command or the time-out value expires (the default is two
minutes).
Once all of the Establish FlashCopy requests have completed, a set of Consistency Group
Created commands must be issued via the same set of DS network interface servers. The
Consistency Group Created commands are directed to each logical subsystem (LSS)
involved in the consistency group. The Consistency Group Created command allows the write
operations to resume to the source volumes.
For a more detailed discussion of the concept of data consistency and how to manage the
Consistency Group operation you can refer to the redbooks IBM System Storage DS8000
Series: Copy Services in Open Environments, SG24-6788 and IBM System Storage DS8000
Series: Copy Services with System z servers, SG24-6787.
Wait FlashCopy
Server2 Wait FlashCopy
FlashCopy
write requests Wait
LSS12
LSS12
LSS12
Wait FlashCopy
Wait
Wait FlashCopy
Waiting write operation
FlashCopy
until Consistency Group Created
command is invoked. Consistency Group
Figure 7-5 Consistency Group FlashCopy
Important: Consistency Group FlashCopy can create host-based consistent copies, they
are not application-based consistent copies. The copies have power-fail or crash level
consistency. This means that if you suddenly power off your server without stopping your
applications and without destaging the data in the file cache, the data in the file cache may
be lost and you may need recovery procedures to restart your applications. To start your
system with Consistency Group FlashCopy target volumes, you may need the same
operations as the crash recovery.
For example, If the Consistency Group source volumes are used with a journaled file
system (like AIX JFS) and the source LUNs are not unmounted before running FlashCopy,
it is likely that fsck will have to be run on the target volumes.
Note: You cannot FlashCopy from a source to a target, where the target is also a Global
Mirror primary volume.
Metro Mirror and Global Copy are explained in 7.3, “Remote mirror and copy” on page 115.
FlashCopy
Source Volume
FlashCopy to
MM or GC
FlashCopy Primary
Target Volume
Volume Secondary
Volume
Primary Metro Mirror
Volume or
Global Copy
Figure 7-6 Establish FlashCopy on existing Metro Mirror or Global Copy primary
Persistent FlashCopy
Persistent FlashCopy allows the FlashCopy relationship to remain even after the copy
operation completes. You must explicitly delete the relationship.
Note: This function is only available through the use of the DS CLI commands and not the
DS Storage Manager GUI.
The remote mirror and copy functions are optional licensed functions of the DS8000, that
include:
Metro Mirror
Global Copy
Global Mirror
Metro/Global Mirror
In the following sections we discuss these remote mirror and copy functions.
For a more detailed and extensive discussion on these topics, you can refer to the redbooks:
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788, and
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787.
Licensing requirements
To use any of these remote mirror and copy optional licensed functions, you must have the
corresponding licensed function indicator feature in the DS8000, and you must acquire the
corresponding DS8000 function authorization with the adequate feature number —license in
terms of physical capacity. For details on features and functions requirements, see
Section 11.1, “DS8000 licensed functions” on page 198.
Also consider that some of the remote mirror solutions, like Global Mirror, or Metro/Global
Mirror, or z/OS Metro/Global Mirror, integrate more than one licensed function. In this case
you will need to have all of the required licensed functions.
Note: for a detailed explanation of the features involved and the considerations you must
have to order Copy Services licensed functions refer to the announcement letters:
IBM System Storage DS8000 — Function Authorization (IBM 239x or 2244)
IBM System Storage DS8000 Series (IBM 242x or 2107)
Server 4 Write
acknowledge
write
1 Write hit
on secondary
3
2
Primary Write to Secondary
(local) secondary (remote)
Figure 7-7 Metro Mirror basic operation
Global Copy does not keep the sequence of write operations. Therefore, the copy is normally
fuzzy, but you can make a consistent copy through synchronization (called a go-to-sync
operation). After the synchronization, you can issue FlashCopy at the secondary site to make
the backup copy with data consistency. After the establish of the FlashCopy, you can change
the mode back to the non-synchronous mode.
Server 2 Write
write acknowledge
Write to secondary
(non-synchronously)
Primary Secondary
(local) (remote)
Figure 7-8 Global Copy basic operation
A consistent and restartable copy of the data at the remote site, created with minimal
impact to applications at the local site.
Data currency where, for many environments, the remote site lags behind the local site
typically 3 to 5 seconds, minimizing the amount of data exposure in the event of an
unplanned outage. The actual lag in data currency that you experience will depend upon a
number of factors, including specific workload characteristics and bandwidth between the
local and remote sites.
Dynamic selection of the desired recovery point objective, based upon business
requirements and optimization of available bandwidth.
Session support whereby data consistency at the remote site is internally managed across
up to eight storage units that are located across the local and remote sites.
Efficient synchronization of the local and remote sites with support for failover and failback
operations, helping to reduce the time that is required to switch back to the local site after
a planned or unplanned outage.
2 Write
Server acknowledge
write
1
Write to secondary
(non-synchronously) FlashCopy
B (automatically)
A
You can see in Figure 7-10 that he A volumes at the local site are the production volumes and
are used as Global Copy primary volumes. The data from the A volumes is replicated to the B
volumes, which are the Global Copy secondary volumes. At a certain point in time, a
Consistency Group is created using all of the A volumes, even if they are located in different
storage units. This has no application impact because the creation of the Consistency Group
is very quick (on the order of milliseconds).
Global Copy
Global Copy FlashCopy
Secondary
Primary Target
FlashCopy Source
FlashCopy
Global Copy
A B C
Once the Consistency Group is created, the application writes can continue updating the A
volumes. The increment of the consistent data is sent to the B volumes using the existing
Global Copy relationships. Once the data reaches the B volumes, it is FlashCopied to the C
volumes.
The C volumes now contain a consistent copy of the data. Because the B volumes except at
the moment of doing the FlashCopy usually contain a fuzzy copy of the data, the C volumes
are used to hold the last consistent point-in-time copy of the data while the B volumes are
being updated by Global Copy.
The data at the remote site is current within 3 to 5 seconds, but this recovery point (RPO)
depends on the workload and bandwidth available to the remote site.
With its efficient and autonomic implementation, Global Mirror is a solution for disaster
recovery implementations where a consistent copy of the data needs to be kept at all times at
a remote location that can be separated by a very long distance from the production site.
Server or Servers
***
normal application I/Os failover application I/Os FlashCopy
Global Copy incremental
asynchronous NOCOPY
Metro Mirror long distance
A B C
Metro Mirror D
synchronous
short distance
Global Mirror
Intermediate site Remote site
Local site (site A) (site B) (siteC)
Figure 7-11 Metro/Global Mirror elements
Both Metro Mirror and Global Mirror are well established replication solutions. Metro/Global
Mirror combines Metro Mirror and Global Mirror to incorporate the best features of the two
solutions:
Metro Mirror
– Synchronous operation supports zero data loss.
– The opportunity to locate the intermediate site disk subsystems close to the local site
allows use of intermediate site disk subsystems in a high availability configuration.
Note: Metro Mirror can be used for distances of up to 300 km but, when used in a
Metro/Global Mirror implementation, a shorter distance might be more appropriate in
support of the high availability functionality.
Global Mirror
– Asynchronous operation supports long distance replication for disaster recovery.
– Global Mirror methodology allows for no impact to applications at the local site.
– Provides a recoverable, restartable, consistent image at the remote site with a
Recovery Point Objective (RPO) typically in the 3-5 second range.
Server or Servers
***
4
normal application I/Os failover application I/Os
Global Mirror network Global Mirror FlashCopy
asynchronous incremental
Metro Mirror large distance NOCOPY
1
2
A B C
3 b
a c
D
Metro Mirror network
synchronous
Global Mirror
small distance
The local site (site A) to intermediate site (site B) component is identical to Metro Mirror.
Application writes are synchronously copied to the intermediate site before write complete is
signaled to the application. All writes to the local site volumes in the mirror are treated in
exactly the same way.
The intermediate site (site B) to remote site (site C) component is identical to Global Mirror,
except that:
The writes to intermediate site volumes are Metro Mirror secondary writes and not
application primary writes.
The intermediate site volumes are both GM source and MM target at the same time.
The intermediate site disk subsystems are collectively paused by the Global Mirror Master
disk subsystem to create the Consistency Group (CG) set of updates. This pause would
normally take 3 ms every 3 to 5 seconds. After the CG set is formed, the Metro Mirror writes
from local site (site A) volumes to intermediate site (site B) volumes, are allowed to continue.
Also, the CG updates continue to drain to the remote site (site C) volumes. The intermediate
site to remote site drain should take only a few seconds to complete.
Once all updates are drained to the remote site, all changes since the last FlashCopy from
the C volumes to the D volumes are logically (NOCOPY) FlashCopied to the D volumes. After
the logical FlashCopy is complete, the intermediate site to remote site Global Copy data
transfer is resumed until the next formation of a Global Mirror CG. The process described
above is repeated every 3 to 5 seconds if the interval for consistency group formation is set to
zero. Otherwise it will be repeated at the specified interval plus 3 to 5 seconds.
The Global Mirror processes are discussed in greater detail in IBM System Storage DS8000
Series: Copy Services in Open Environments, SG24-6788 and IBM System Storage DS8000
Series: Copy Services with System z servers, SG24-6787.
Figure 7-13 illustrates the basic operations characteristics of z/OS Global Mirror.
Primary Secondary
SDM manages the
site site
data consistency System
Data
Write Mover
acknowledge
Server 2
write
1 Write
ashynchronously
The SDM can be located at the secondary site, at the primary site, or at an independent site.
Figure 7-14 on page 123 illustrates the basic operation characteristics of a z/OS Metro/Global
Mirror implementation.
SDM
Metropolitan Unlimited
distance distance
Metro
Mirror z/OS Global
Mirror
P X’
P’ X
DS8000 DS8000 DS8000 X”
Metro Mirror Metro Mirror/ z/OS Global Mirror
Secondary z/OS Global Mirror Secondary
FlashCopy
Primary
Metro Mirror
Metro Mirror is a function for synchronous data copy at a distance. The following
considerations apply:
There is no data loss and it allows for rapid recovery for distances up to 300 km.
There may be slight performance impact for write operations.
Global Copy
Global Copy is a function for non-synchronous data copy at very long distances —only limited
by the network implementation. The following considerations apply:
It can copy your data at nearly an unlimited distance, making it suitable for data migration
and daily backup to a remote distant site.
The copy is normally fuzzy but can be made consistent through a synchronization
procedure.
To create a consistent copy for Global Copy, you need a go-to-sync operation —that is,
synchronize the secondary volumes to the primary volumes. During the go-to-sync operation,
the mode of remote copy changes from a non-synchronous copy to a synchronous copy.
Therefore, the go-to-sync operation may cause performance impact to your application
system. If the data is heavily updated and the network bandwidth for remote copy is limited,
the time for the go-to-sync operation becomes longer.
Global Mirror
Global Mirror is an asynchronous copy technique; you can create a consistent copy in the
secondary site with an adaptable Recovery Point Objective (RPO) —Recovery Point
Objective (RPO) specifies how much data you can afford to re-create should the system need
to be recovered. The following considerations apply:
Global Mirror can copy to nearly an unlimited distance.
It is scalable across the storage units.
It can realize a low RPO if there is enough link bandwidth —when the link bandwidth
capability is exceeded with a heavy workload, the RPO might grow.
Global Mirror causes only a slight impact to your application system.
Note: Not always all the Copy Services options are available in each management
interface. Refer to the following redbooks for specific considerations:
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787
The network components for Copy Services are illustrated in Figure 7-15 on page 125.
S-HMC 1 Processor
Ethernet
(internal) Complex 0
Switch 1
Customer DS8000
Network
DS Storage
Manager Ethernet Processor
(S-HMC 2)
Switch 2 Complex 1
DS CLI (external)
DS API
Each DS8000 will have an internal S-HMC in the base frame, and you can have an external
S-HMC for redundancy.
For further information about the S-HMC, see Chapter 9, “DS HMC planning and setup” on
page 145.
DS Storage Manager is already installed in the S-HMC. You can also install it in other
computers that you may set up. When you manage the Copy Services functions with DS SM
in computers that you set up, DS SM issues its command to the S-HMC via the Ethernet
network.
The DS Storage Manager supports almost all the Copy Services options. The following
options are not supported:
Consistency Group operation —for FlashCopy and for Metro Mirror
Inband commands over remote mirror links
For more information on the DS SM Web-based graphical user interface you can refer to the
following sources:
IBM System Storage DS8000 Information Center web site at:
http://publib.boulder.ibm.com/infocenter/dsichelp/ds8000ic/index.jsp
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787
For additional information about the DS SM uses see also Chapter 13, “Configuration with DS
Storage Manager GUI” on page 223.
For more information on the DS CLI, as well as the supported platforms you can refer to the
following sources:
IBM System Storage DS8000 Information Center web site at:
http://publib.boulder.ibm.com/infocenter/dsichelp/ds8000ic/index.jsp
DS 8000 Interoperability matrix, that you can find at:
http://www-03.ibm.com/servers/storage/disk/ds8000/pdf/ds8000-matrix.pdf
IBM System Storage DS8000: Command-Line Interface User´s Guide, SC26-7916.
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787
For additional information about the DS CLI uses, see also Chapter 14, “Configuration with
Command Line interface” on page 267.
TPC for Replication is designed to help administrators manage Copy Services. This applies
not only to the copy services provided by DS8000 and DS6000 but also to copy services
provided by the ESS 800 and SAN Volume Controller (SVC).
In addition to these capabilities, TPC for Replication also provides two-site Business
Continuity manageability. This is intended to provide disaster recovery management through
planned and unplanned failover and failback automation, and monitoring progress of the copy
services so you can verify the amount of replication that has been done as well as the amount
of time required to complete the replication operation.
For more information about TPC for replication, you can refer to the following sources:
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787
IBM TotalStorage Productivity Center for Replication User’s Guide, SC32-0103
IBM TotalStorage Productivity Center for Replication Installation and Configuration Guide,
SC32-0102
IBM TotalStorage Productivity Center for Replication Command-Line Interface User’s
Guide, SC32-0104.
volume spaces. The DS Open API also enables Copy Services functions such as FlashCopy
and remote mirror and copy functions. It supports these activities through the use of the
Storage Management Initiative Specification (SMIS), as defined by the Storage Networking
Industry Association (SNIA)
The DS Open API helps integrate DS configuration management support into storage
resource management (SRM) applications, which allow customers to benefit from existing
SRM applications and infrastructures. The DS Open API also enables the automation of
configuration management through customer-written applications. Either way, the DS Open
API presents another option for managing storage units by complementing the use of the IBM
System Storage DS Storage Manager web-based interface and the DS Command-Line
Interface.
You must implement the DS Open API through the IBM System Storage Common Information
Model (CIM) agent, a middleware application that provides a CIM-compliant interface. The DS
Open API uses the CIM technology to manage proprietary devices such as open system
devices through storage management applications. The DS Open API allows these storage
management applications to communicate with a storage unit.
For information on the DS Open API, refer to the publication IBM System Storage DS Open
Application Programming Interface Reference, GC35-0516.
These interfaces have the advantage of not having to issue their commands to the DS8000
HMC. They can instead directly send inband commands over a FICON channel connection
between the System z and the DS8000. Sending inband commands allows for a very quick
command transfer that does not depend on any additional software stacks.
7.5 Interoperability
Remote mirror and copy pairs can only be established between disk subsystems of the same
(or similar) type and features. For example, a DS8000 can have a remote mirror pair with
another DS8000, a DS6000, an ESS 800, or an ESS 750. It cannot have a remote mirror pair
with an RVA or an ESS F20. Note that all disk subsystems must have the appropriate features
installed. If your DS8000 is being mirrored to an ESS disk subsystem, the ESS must have
PPRC Version 2 (which supports Fibre Channel links) with the appropriate licensed internal
code level (LIC).
Note: The DS8000 does not support ESCON links for remote mirror and copy operations.
If you want to establish a remote mirror relationship between a DS8000 and an ESS 800,
you have to use FCP links.
Part 2
The publication IBM System Storage DS8000 Introduction and Planning Guide, GC35-0515,
should be reviewed and available to use during the configuration and installation process.
In general the following should be considered for your installation planning checklist:
Plan for growth to minimize disruption to operations. Expansion frames can only be placed
to the right (from the front) of the DS8000
Location suitability, floor loading and access constraints, elevators, doorways, and so on
Power requirements: redundancy, use of Uninterrupted Power Supply (UPS)
Environmental requirements: adequate cooling capacity
A plan detailing the desired logical configuration of the storage
Staff education and availability to implement the storage plan. Alternatively, IBM or IBM
Business Partner services
In this chapter you will find information that will assist you with the planning and installation
activities.
A Storage Administrator should also coordinate requirements from the user applications and
systems in order to build a storage plan for the installation. This will be needed to configure
the storage after the initial hardware installation is complete.
The following people should be briefed and engaged in the planning process for the physical
installation:
The IBM System Storage DS8000 Introduction and Planning Guide, GC35-0515, contains
additional information about physical planning. You can download it at:
http://www.ibm.com/servers/storage/disk/ds8000
Table 8-1 shows the final packaged dimensions and maximum packaged weight of the
DS8000 storage unit shipgroup.
Model 92E expansion unit Height 207.5 cm (81.7 in.) 1209 kg (2665 lb)
Model 9AE expansion unit Width 101.5 cm (40 in.)
pallet or crate Depth 137.5 cm (54.2 in.)
(if ordered) External S-HMC Height 69.0 cm (27.2 in.) 75 kg (165 lb)
container Width 80.0cm (31™.5 in.)
Depth 120.0cm (47.3 in.)
Attention: A fully configured model in the packaging can weigh over 1406 kg (3100 lbs).
Use of less than three persons to move it can result in injury.
The total weight and space requirements of the storage unit will depend on the configuration
features you ordered. You may need to consider calculating the weight of the unit and the
expansion box (if ordered) in their maximum capacity to allow for the addition of new features.
Important: You need to check with the building engineer or another appropriate person to
be sure that the floor loading is properly considered.
Raised floors can better accommodate cabling layout. The power and interface cables enter
the storage unit via the rear side.
Figure 2-1 shows the location of the cable cutouts. You may use the following measurements
when you cut the floor tile:
Width: 45.7 cm (18.0 in.)
Depth: 16 cm (6.3 in.)
The storage unit location area should also cover the service clearance needed by IBM
services representatives when accessing the front and rear of the storage unit. You may use
the following minimum service clearances —the dimensions are also shown in Figure 8-2 on
page 136:
1. For the front of the unit, allow a minimum of 121.9 cm (48 in.) for the service clearance.
2. For the rear of the unit, allow a minimum of 76.2 cm (30 in.) for the service clearance.
3. For the sides of the unit, allow a minimum of 5.1 cm (2 in.) for the service clearance.
Power connectors
Each DS8000 base and expansion unit has redundant power supply systems. The two line
cords to each frame should be supplied by separate AC power distribution systems. Table 8-4
lists the standard connector and receptacle for the DS8000.
Use a 60 Ampere rating for the low voltage feature and a 25 Ampere rating for the high
voltage feature.
For more details regarding power connectors and line cords, refer to the publication IBM
System Storage DS8000 Introduction and Planning Guide, GC35-0515.
Input voltage
The DS8000 supports a three-phase input voltage source. Table 8-5 shows the power
specifications for each feature code.
Nominal input voltage 200, 208, 220, or 240 RMS Vac 380, 400, 415, or 480 RMS Vac
(3-phase)
Peak electric power 5.5 kVA 7.0 kVA 7.0 kVA 6.0 kVA
Air circulation for the DS8000 is provided by the various fans installed throughout the frame.
The power complex and most of the lower part of the machine take air from the front and
exhaust air to the rear. The upper disk drive section takes air from the front and rear sides,
and exhausts air to the top of the machine.
The recommended operating temperature for the DS8000 is between 20 to 25o C (68 to 78o F)
at a relative humidity range of 40 to 50 percent.
Important: Make sure that air circulation for the DS8000 base unit and expansion units is
maintained free from obstruction to keep the unit operating in the specified temperature
range.
For more details regarding power control features, refer to the publication IBM System
Storage DS8000 Introduction and Planning Guide, GC35-0515.
ESCON
The DS8000 ESCON adapter supports two ESCON links per card. Each ESCON port is a
64-bit, LED-type interface, which features an enhanced microprocessor, and supports 62.5
micron multimode fiber optic cable terminated with the industry standard MT-RJ connector.
ESCON cables can be specified when ordering the ESCON host adapters.
Table 8-7 shows the various fiber optic cable features available for the ESCON ports.
The 31-meter cable is the standard length provided with the DS8000. You may order custom
length cables from IBM Global Services.
Note: Feature 1432 is a conversion cable for use in the DS8000 when connecting the unit
to a S/390 host using existing cables. The 9672 processor and the IBM ESCON Director
(9032) use duplex connectors.
Fibre Channel/FICON
The DS8000 Fibre Channel/FICON adapter has four ports per card. Each port supports FCP
or FICON, but not simultaneously. Fabric components from IBM, CNT, McDATA, and Brocade
are supported by both environments.
FCP is supported on point-to-point, fabric, and arbitrated loop topologies. FICON is supported
on point-to-point and fabric topologies.
The 31-meter fiber optic cable can be ordered with each Fibre Channel adapter. Using the
9-micron cable, a longwave adapter can extend the point-to-point distance to 10 km. A
shortwave adapter using 50 micron cable supports point-to-point distances of up to 500
meters at 1 Gbps and up to 300 meters at 2 Gbps. Additional distance can be achieved with
the use of appropriate SAN fabric components.
Table 8-8 lists the various fiber optic cables features for the FCP/FICON adapters.
Note: The remote mirror and copy functions use FCP as the communication link between
DS8000s, DS6000s, and ESS Models 800 and 750.
For more details on IBM-supported attachments, refer to the publication IBM System Storage
DS8000 Host Systems Attachment Guide, SC26-7917.
For the most up-to-date details about host types, models, adapters, and operating systems
supported by the DS8000 unit, see the Interoperability Matrix at:
http://www.ibm.com/servers/storage/disk/ds8000/interop.html
Check your local environment for the following DS8000 unit connections:
Hardware management console network access
DS CLI console
Remote support connection
Remote power control
Storage area network connection
For more details on physical network connectivity, refer to the publications IBM System
Storage DS8000 User´s Guide, SC26-7915 and IBM System Storage DS8000 Introduction
and Planning Guide, GC35-0515.
Tip: To ensure that the IBM service representative can quickly and easily access an
external DS HMC, place the external DS HMC rack within 15.2 m (50 ft) of the storage
units that are connected to it.
The management console can be connected to your network for remote management of your
system using the DS Storage Manager Web-based graphical user interface (GUI), the DS
Command-Line Interface (CLI), or using storage management software through the DS Open
API. In order to use the CLI to manage your storage unit, you need to connect the
management console to your LAN because the CLI interface is not available on the HMC. The
DS8000 can be managed from the HMC only using the Web GUI interface. Connecting the
HMC to your LAN also allows you to access the DS Storage Manager from any location that
has network access using a Web browser.
In order to connect the management consoles (internal, and external if present) to your
network, you need to provide the following settings to your IBM service representative so that
they can configure the management consoles for attachment to your LAN:
Management console network IDs, host names, and domain name
Dynamic Name Server settings (if you plan to use DNS to resolve network names)
Routing information
For details on the hardware and software requirements for the DS CLI, refer to the IBM
System Storage DS8000: Command-Line Interface User´s Guide, SC26-7916.
You can take advantage of the DS8000 remote support feature for outbound calls (Call Home
function) or inbound calls (remote service access by an IBM support representative). You
need to provide an analog telephone line for the DS HMC modem.
Take note of the following guidelines to assist in the preparation for attaching the DS8000 to
the client’s LAN:
1. Assign a TCP/IP address and host name to the DS HMC in the DS8000.
2. If e-mail notification of service alert is allowed, enable the support on the mail server for
the TCP/IP addresses assigned to the DS8000.
3. Use the information that was entered on the installation worksheets during your planning.
IBM recommends service connection through the high-speed VPN network utilizing a secure
Internet connection. You need to provide the network parameters for your DS HMC through
the installation worksheet prior to actual configuration of the console. See Chapter 9, “DS
HMC planning and setup” on page 145.
Your IBM System Support Representative (SSR) will need the configuration worksheet during
the configuration of your DS HMC. A worksheet is available in the IBM System Storage
DS8000 Introduction and Planning Guide, GC35-0515.
See also Chapter 20, “Remote support” on page 391 for further discussion on remote support
connection.
In a System z environment, the host must have the Power Sequence Controller (PSC) feature
installed to have the ability to turn on/off specific control units, like the DS8000. The control
unit is controlled by the host through the power control cable. The power control cable comes
with a standard length of 31 meters, so be sure to consider the physical distance between the
host and DS8000.
A SAN allows your single Fibre Channel host port to have physical access to multiple Fibre
Channel ports on the storage unit. You may need to establish zones to limit the access (and
provide access security) of host ports to your storage ports. Take note that shared access to a
storage unit Fibre Channel port might come from hosts that support a combination of bus
adapter types and operating systems.
Note: The DS8000 storage unit does not support ESCON links for the remote mirror and
copy functions —only Fibre Channel links for remote mirror and copy connectivity.
Make sure that you have a sufficient number of FCP paths assigned for your remote mirroring
between your source and target sites to address performance and redundancy issues. When
you plan to use both Metro Mirror and Global Copy modes between a pair of storage units,
IBM recommends that you use separate logical and physical paths for the Metro Mirror and
another set of logical and physical paths for the Global Copy.
Plan accordingly on the distance between the primary and secondary storage units to
properly acquire the necessary length of fiber optic cables you need, or if your Copy Services
solution would require separate hardware such as channel extenders or DWDM.
For detailed information refer to the redbooks IBM System Storage DS8000 Series: Copy
Services in Open Environments, SG24-6788 and IBM System Storage DS8000 Series: Copy
Services with System z servers, SG24-6787.
A minimum of four spares of the largest capacity array site on the DA pair
A minimum of two spares of capacity and rpm greater than or equal to the fastest array
site of any given capacity on the DA pair
The DDM sparing policies support the over configuration of spares. This possibility may be of
interest to some installations as it allows the repair of some DDM failures to be deferred until
a later repair action is required.
Refer to IBM System Storage DS8000 Introduction and Planning Guide, GC35-0515 for more
details on the DS8000 sparing concepts. See also “Spare creation” on page 85.
Table 8-9 helps you plan for the capacity of your DS8000 system.
Table 8-9 Disk drive set capacity for open systems and System z environments
Disk size Physical Rank type Effective capacity of one Rank in decimal GB
(GB) capacity (Number of extents)
per disk
drive set in Rank of RAID 10 arrays Rank of RAID 5 arrays
decimal GB
3+3 4+4 6+P 7+P
Note:
1. Physical capacities are in decimal gigabytes (GB). One GB is one billion bytes.
2. Please keep in mind the lower recommended usage for the 500 GB FATA drives as
detailed in Section 4.4.6, “FATA versus Fiber Channel drives on the DS8000” on
page 62
Table 8-9 shows the effective capacity of one rank in the different possible configurations. A
disk drive set contains 16 drives, which form two array sites. Capacity on the DS8000 is
added in increments of one disk drive set. The effective capacities in the table are expressed
in binary gigabytes and as the number of extents.
These new drives are meant to complement, not compete with existing Fibre Channel drives
as they are not intended for use in applications that require drive utilization duty cycles
greater than 20 percent. Intermix of the two drive types within the DS8000 is supported with
certain restrictions on physical placement and logical configuration.
For a detailed discussion on FATA drives you can refer to Chapter 4 Hardware Components,
sections 4.4.4 on page 56, 4.4.5 on page 58, and 4.4.6 on page 62.
Planning for future growth would normally suggest an increase in physical requirements, in
your installation area (including floor loading), electrical power, and environmental cooling.
A key feature that you can order for your dynamic storage requirement is the Standby
Capacity on Demand (CoD). This offering is designed to provide you with the ability to tap into
additional storage and is particularly attractive if you have rapid or unpredictable growth, or if
you simply want the knowledge that the extra storage will be there when you need it. Standby
Cod allows you to access the extra storage capacity when you need it through a nondisruptive
activity. For more information on Capacity on Demand, see Section 21.2, “Using Capacity on
Demand (CoD)” on page 400.
The DS HMC is connected to the storage facility by way of redundant private Ethernet
networks. Figure 9-1 shows the back of a single DS HMC and a pair of Ethernet switches.
SFI-1 SFI-2
Port for
Customer
Network
SFI-1 SFI-2
Figure 9-1 Rear view of the DS HMC and a pair of Ethernet switches
The DS HMC has two built-in Ethernet ports: One dual-port Ethernet PCI adapter and one
PCI modem for asynchronous Call Home support. The DS HMCs private Ethernet ports
shown are configured into port 1 of each Ethernet switch to form the private DS8000 network.
The client Ethernet port indicated is the primary port to be used to connect to the client
network. The empty Ethernet port is normally not used. Corresponding private Ethernet ports
of the external DS HMC (FC1110) would be plugged into port 2 of the switches, as shown. To
interconnect two DS8000 base frames, FC1190 would provide a pair of 31m Ethernet cables
to connect from port 16 of each switch in the second base frame into port 15 of the first frame.
If the second DS HMC is installed in the second DS8000, it would remain plugged into port 1
of its Ethernet switches. Each LPAR of a Storage Facility Image (SFI) is connected via a
redundant Ethernet connection to the internal network.
1
A processor LPAR is part of a Storage Facility Image. For the client there is no possibility to manage a processor
LPAR, as is possible on the System p5 or System i5 systems.
Figure 9-2 shows the management interface components and the way they communicate.
The GUI and the CLI are comprehensive, easy-to-use interfaces for a storage administrator to
perform DS8000 management tasks. They can be used interchangeably depending on the
particular task. The DS Open API provides an interface for storage management application
programs to the DS8000.
Installable code for the interface programs is available on CDs that are delivered with the
DS8000 unit. Subsequent versions are made available with DS8000 Licensed Internal Code
updates. All front ends for the DS8000 can be installed on any of the supported workstations.
Tip: We recommend that you have a directory structure in place where all software
components to be installed for the DS environment are stored, including the latest levels
from the Internet used for installation.
The DS Storage Management Server (DS SMS) runs in a WebSphere environment installed
in the DS HMC. The DS SMS provides the communication interface to the front end DS
Storage Manager (DS SM), which is running in a Web browser. The DS SMS also
communicates with the DS Network Interface Server (DS NW IFS), which is responsible for
communication with the two controllers of the DS8000.
DS Storage Manager can also be accessed locally from the DS8000 Management console
using the Web browser that comes pre installed on the DS HMC. See “Using the DS GUI on
the HMC” on page 150 for instructions on how to log on to the Management console and to
access the DS Storage Manager.
Note: The DS CLI cannot be used locally on the DS Hardware Management console.
Once the DS CLI has been installed on a workstation it can be used by just typing dscli in the
command environment. Multiple DS CLI commands can be integrated into a script that can be
executed by using the dscli command with the -script parameter.
To enter the interactive mode of the DS CLI just type dscli in a command prompt window and
follow the prompts to log on, as shown in Example 9-2. Once logged on, DS CLI commands
can be entered one at a time.
To call a script with DS CLI commands, the following syntax in a command prompt window of
a Windows workstation can be used:
dscli -script <script_filename> -hmc1 <ip-address> -user <userid> -passwd <password>
In Example 9-3, script file lssi.cli contains just one CLI command, the lssi command.
Example 9-4 shows how to run a single command from a workstation prompt. The -cfg flag
precedes a CLI profile that contains the IP address and user ID information for the DS HMC.
The new Common Information Model (CIM) Agent (v5.2) for the DS8000 is Storage
Management Initiative Specification (SMI-S) 1.1 compliant. This agent, previously also known
as the Installation and Configuration Automation/Assistance Tool (ICAT), is used by storage
management applications such as TPC, TSM, VSS/VDS, and Director & Director Server
Storage Provisioning. Also, with compliance to more open standards, the agent can now be
accessed by software from third party vendors including Veritas/Symantec, HP/AppIQ, EMC,
and many other applications at SNIA Interoperability Lab.
Starting with release v5.2.1, it is possible to install the CIM agent directly on the DS8000
HMC thus eliminating the need for a customer supplied server. However, with the HMC based
implementation, it is only possible to access the DS8000 managed by that HMC. For this
reason, users may opt to install it on an external system, allowing them to manage multiple
devices of different types (DS8000, DS6000 or ESS 800/Fxx).
Installing the agent directly in the DS8000 HMC has a few other limitations, including:
Limited failover between primary and secondary HMC
Must use WEB SM to enable the agent
Must use DSCIMCLI to configure the agent
No integration with DS8000 GUI or DSCLI
The CIM agent is bundled with MM Extensions CD and can be loaded using the CDA
procedure. After being loaded, the agent must be enabled using the WEB SM and configured
with the DSCIMCLI.
3. Click Browser. The Web browser is started with no address bar and a Web page titled
WELCOME TO THE DS8000 MANAGEMENT CONSOLE is displayed; see Figure 9-4.
– Password: admin
The password must be changed on first login. If someone has already logged on, check
with the person what the new password is.
7. A Wand (password manager) panel opens. Select OK.
The typical setup for a DS HMC environment assumes that the following servers and
workstations exist. See Figure 9-5.
DS Hardware Management Console (online configuration mode only)
One Management console, with all the required software pre installed, is always located
within the first DS8000 unit ordered. This can be used for a real-time (online)
configuration. Optionally, a second Management console can be ordered and installed in a
client provided rack or in a second DS8000.
DS SM workstations with DS SM code installed (online/offline-configuration modes)
Client-provided workstation where DS Storage Manager has been installed. This can be
used for both real-time (online) and simulated (offline) configurations.
Workstations with Web browser only (online configuration mode only)
User-provided workstations with a Web browser to access the DS Storage Manager on
the DS HMC. This can be used for a real-time (online) configuration.
Note: An implementation with only one DS HMC should only be used where high
availability of the storage environment is not needed.
Note: The DS CLI is usually a better option for mass updates due to web page load times.
Figure 9-6 shows the Welcome panel of the DS8000 Storage Manager. The main areas that
can be handled using the DS SM are:
Monitoring and administration
Hardware management
Storage configuration
Copy Services management
A Windows or Linux version of the DS HMC can be used to simulate configuration tasks in
offline mode. This option is not available with the internal or external DS HMCs.
Also, it is easy to integrate the CLI commands into existing scripts. This might, for example,
be needed where automation for backup and disaster recovery is running based on scripts.
Note: The DS CLI consists of commands to interact with the DS8000. Multiple commands
can be integrated in one DS CLI script. Programming logic needs to be implemented in the
software that uses the DS CLI scripts or DS CLI commands.
The Web browser is the only software that is needed on workstations that will do
configuration tasks online using the DS Storage Manager GUI.
Also, depending on your Metro/Global Mirror configuration, there are the Metro Mirror Add on,
and the Global Mirror Add on licensed functions for the Turbo models. For the 921/922/9A2
models Metro Mirror and Global Mirror are under the Remote Mirror and Copy (RMC)
licensed function, and the FICON/ESCON Attachment function does not apply.
To prepare for the activation of the license keys, the Disk storage feature activation (DSFA)
Internet page can be used to create activation codes and to download keyfiles, which then
need to be applied to the DS8000. The following information is needed for creating license
activation codes:
Serial number of the DS8000
The serial number of a DS8000 can be taken from the front of the base frame (lower right
corner). If several machines have been delivered this is the only way to obtain the serial
number of a machine located in a specific point in the computer center.
Machine signature
The machine signature can only be obtained using the DS SM or the DS CLI after
installation of the DS8000 and DS HMC.
License Function Authorization document
In most situations, the DSFA application can locate your 239x/2244 function authorization
record when you enter the DS8000 (242x/2107) serial number and signature. However, if
the function authorization record is not attached to the 242x/2107 record, you must assign
it to the 242x/2107 record in the DSFA application. In this situation, you will need the
239x/2244 serial number (which you can find on the License Function Authorization
document).
If you are activating codes for a new storage unit, the License Function Authorization
documents are included in the shipment of the storage unit. If you are activating codes for
an existing storage unit, IBM sends these documents to you in an envelope.
For more information on required DS8000 features and function authorizations, as well as
activation tasks, see Chapter 11, “Features and license keys” on page 197.
The Operating Environment license must be for a capacity greater than or equal to the total
physical capacity of the system. If it is not, you will not be able to configure any storage for a
new box or the new capacity for an upgrade. For each of the other features you need to order
a license for capacity greater than or equal to the capacity of the storage format with which it
will be used. For example, assume you have a 10 TB box with 4 TB of storage for CKD and 6
TB for FB. If you only wanted to use Metro Mirror for the CKD storage you would need to order
the Metro Mirror license for 4 TB.
Note: Applying increased feature activation codes is a concurrent action, but a license
reduction or deactivation is a disruptive action.
During planning the capacity ordered for each of the Copy Services functions should be
reviewed. After activation of the features it should be checked that they match the capacity
assigned in the DS8000 for Copy Services functions.
Important: The Interoperability Matrix always reflects information regarding the latest
supported code levels. This does not necessarily mean that former levels of HBA
firmwares or drivers are no longer supported. If in doubt about any supported levels
contact your IBM representative.
The Interoperability Matrix and the HBA Support Matrix can respectively be found at the
IBM System Storage technical support Web site at:
http://www-03.ibm.com/servers/storage/disk/ds8000/pdf/ds8000-matrix.pdf
http://www-03.ibm.com/servers/storage/support/config/hba/index.wss
Maintenance windows
Even though the microcode update of the DS8000 is a non-disruptive action, any
prerequisites identified for the hosts (for example, patches, new maintenance levels, new
drivers) could make it necessary to schedule a maintenance window. The host
environments can then be upgraded to the level needed in parallel to the microcode
update of the DS8000 taking place.
For more information on microcode upgrades see Chapter 18, “Licensed machine code” on
page 375.
The setting of the date and time of the DS8000 is a maintenance activity that can only be
done by the IBM CE.
Call Home is the capability of the DS HMC to contact IBM support to report a serviceable
event. Remote Services is the capability of IBM service representatives to connect to the DS
HMC to perform service tasks remotely. If allowed to do so by the setup of the client’s
environment, an IBM Product Field Engineer (IBM PFE) could connect to the DS HMC to
perform detailed problem analysis. He can view error logs and problem logs, and initiate trace
or dump retrievals.
Remote support can be configured for dial-up connection through a modem or high-speed
virtual private network (VPN) Internet connection. The details to set up the remote support
functionality are described in the VPN Security and Implementation document, which can be
found on the DS8100 Support Web page under topic VPN Implementation:
http://www.ibm.com/servers/storage/support/disk/ds8100/installing.html
When the VPN connection is used, if there is a firewall in place to shield your network from
the open Internet, the firewall must be configured to allow the HMC to connect to the IBM
VPN servers. The HMC establishes connection to the following TCP/IP addresses:
207.25.252.196 IBM Boulder VPN Server
129.42.160.16 IBM Rochester VPN Server
The setup of the remote support environment is done by the IBM CE during initial installation.
Attention: The password of the admin user ID will need to be changed before it can be
used. The GUI will force you to change the password when you first log in. The DS CLI will
allow you to log in, but will not allow you to issue any other commands until you have
changed the password. As an example, to change the admin user’s password to passw0rd,
use the following command: chuser-pw passw0rd admin. Once you have issued that
command, you can then issue any other command.
During the planning phase of the project, a worksheet or a script file was established with a
list of all people who need access to the DS GUI or DS CLI. The supported roles are:
Administrator has access to all available commands.
Physical operator has access to maintain the physical configuration (storage complex,
storage image, array, rank, and so on).
Logical operator has access to maintain the logical configuration (logical volume, host,
host ports, and so on).
Copy Services operator has access to all Copy Services functions and the same access
as the monitor group.
Monitor group has access to all read-only list and show commands.
No access could be used by the administrator to temporarily deactivate a user ID.
General password settings include the time period in days after which passwords expire and a
number that identifies how many failed logins are allowed.
Whenever a user is added, a password is entered by the administrator. During the first sign-in,
this password needs to be changed by the user. The user ID is deactivated if an invalid
password is entered more times then as defined by the administrator for the password
settings. Only a user with administrator rights can then reset the user ID with a new initial
password.
If the access is denied for the administrator due to the number of invalid tries, a procedure
can be obtained from your IBM representative to reset the administrator’s password.
Tip: User names and passwords are both case sensitive. If you create a user name called
Anthony, you cannot log on using the user name anthony. DS CLI commands, however, are
not case sensitive. So the commands LSUSER or LSuser or lsuser will all work.
The password for each user account is forced to adhere to the following rules:
The length of the password must be between six and 16 characters.
It must begin and end with a letter.
It must have at least five letters.
It must contain at least one number.
It cannot be identical to the user ID.
It cannot be a previous password.
rmuser
This command removes an existing user ID. In Example 9-6 we remove a user called
Enzio.
chuser
This command changes the password and/or group of an existing user ID. It is also used
to unlock a user ID that has been locked by exceeding the allowable login retry count. You
could also lock a user ID if desired. In Example 9-7 on page 160 we unlock the user,
change the password, and change the group membership for a user called Sharon. A user
has to use the chpass command when they use that user ID for the first time.
lsuser
With this command, a list of all user IDs can be generated. In Example 9-8 we can see
three users.
showuser
The account details of user IDs can be displayed with this command. In Example 9-9 we
list the details of Arielle’s user ID.
Example 9-9 Using the show user command to list user information
dscli> showuser Arielle
Date/Time: 10 November 2005 21:25:34 IBM DSCLI Version: 5.1.0.204
Name Arielle
Group op_volume
State active
FailedLogin 0
managepwfile
This command creates or adds to an encrypted password file that will be placed onto the
local machine. This file can be referred to in a DS CLI profile. This allows you to run scripts
without specifying a DS CLI user password in clear text. If manually starting DS CLI you
can also refer to a password file with the -pwfile parameter. By default, the file is placed
in: c:\Documents and settings\<Windows user name>\DSCLI\security.dat or in
$HOME/dscli/security.dat (for non Windows based operating systems)
In Example 9-10 we manage our password file by adding the user ID called BenColeman.
The password is now saved in an encrypted file called security.dat.
chpass
This command lets you change two password rules: password expiration (days) and failed
logins allowed. In Example 9-11 on page 161 we change the expiration to 365 days and 5
failed logon attempts. If you set both values to zero, then passwords never expire and
unlimited logon attempts are allowed —which is not recommended.
showpass
This command lists the properties for passwords (Password expiration days and Failed
logins allowed). In Example 9-12 we can see that passwords have been set to expire in
365 days and that 5 login attempts are allowed before a user ID is locked.
The exact syntax for any DS CLI command can be found in the IBM System Storage DS8000:
Command-Line Interface User´s Guide, SC26-7916. You can also use the DS CLI help
command to get further assistance.
The administrator can perform several tasks from the Select Action pull-down:
Add User (DS CLI equivalent is mkuser)
Modify User (DS CLI equivalent is chuser)
Lock or Unlock User - Choice will toggle based on user state (DS CLI equivalent is chuser)
Delete User (DS CLI equivalent is rmuser)
Password Settings (DS CLI equivalent is chpass)
If you click a user name, it will bring up the Modify User panel.
Note: If a user who is not in the Administrator group logs onto the DS GUI and goes to the
User Administration panel, they will only be able to see their own user ID in the list. The
only action they will be able to perform is to change their password.
Selecting Add User will display a window in which a user can be added by entering the user
ID, the temporary password, and the role (see Figure 9-8). The role will decide what type of
activities can be performed by this user. In this window, the user ID can also be temporarily
deactivated by selecting Access none.
In Figure 9-8, a user is being added with the user name Frontdesk. This user is being placed
into the Monitor group.
The external DS HMC is an optional priced feature. To help preserve console functionality, the
external and the internal DS HMCs are not available to be used as a general purpose
computing resource.
In the next few sections we use the terms first and internal DS HMC interchangeably, and
second and external DS HMC interchangeably. The two DS HMCs run in a dual-active
configuration, so either DS HMC can be used at any time. The distinction between the
internal and external unit is only for clarity and ease of explanation.
Tip: To ensure that IBM service representatives can quickly and easily access an
external DS HMC, place its rack within 15.2 m (50 ft) of the storage units it connects.
In order to take advantage of the high availability features of the second HMC (high availability
for configuration operations and Advanced Copy Services), you must configure the DS CLI or
the DS SM to take advantage of the second HMC.
When the user issues a configuration or copy services command, the DS CLI or DS SM will
send the command to the first HMC. If the first HMC is not available, it will automatically send
the command to the second HMC instead. Typically, the user does not have to re-issue the
command.
Any changes made using the first HMC are instantly reflected in the second HMC. There is no
caching done within the HMC, so there are no cache coherency issues. By first HMC, we
mean the HMC defined as the HMC1. It is even possible to define the external HMC as the
first HMC, and vice versa, but this is not typical.
Alternatively, you can modify the following lines in the dscli.profile (or other profile) file:
# Management Console/Node IP Address(es)
# hmc1 and hmc2 are equivalent to -hmc1 and -hmc2 command options.
hmc1:mitmuzik.ibm.com
hmc2:mitgasse.ibm.com
After you make these changes, the DS CLI will use the second HMC in the unlikely event that
the first HMC fails. This change will allow you to perform both configuration and copy services
commands with full redundancy.
One alternative is to install the DS Storage Manager on a separate system. In this way, even
if the internal HMC fails, the separate system will be able to continue your configuration or
copy services session by connecting to the external HMC.
The DS Storage Manager can be configured to use a second HMC by specifying both the first
and second HMC in the Real-time manager → Manage hardware → Storage complexes
screen. From the pull-down menu, choose Add. See Figure 9-9.
On the next panel, check the Define Management console 2 check box, and add the IP
addresses of both HMC machines. Then click OK. Figure 9-10 on page 165 shows the adding
of two HMCs, 10.10.10.10 and 10.10.10.11.
At this point, the DS Storage Manager is configured to use the second HMC if the first HMC
should fail.
10
There are some shortcomings with FC-AL. The most obvious ones are:
Arbitration - disks compete for loop bandwidth.
Failures within the FC-AL loop, particularly with intermittently failing components on the
loops and disks.
Increased time it to complete a loop operation as the number of loop devices increase.
For highly parallel operations, concurrent reads and writes with various transfer sizes, this
would impact the total effective bandwidth of an FC-AL structure.
To host servers
Adapter Adapter
Adapter Adapter
Memory
Storage server
Processor Processor
To storage servers
Adapter Adapter
20 port switch
o oo
16 DDM
20 port switch
These switches use FC-AL protocol and attach FC-AL drives through a point-to-point
connection. The arbitration message of a drive is captured in the switch, processed and
propagated back to the drive, without routing it through all the other drives in the loop.
Performance is enhanced since both DAs connect to the switched Fibre Channel disk
subsystem back end as displayed in Figure 10-2. Note that each DA port can concurrently
send and receive data.
To host servers
Adapter Adapter
Storage server
Adapter Adapter
To next
switch
20 port switch
ooo
16 DDM
20 port switch
Figure 10-2 High availability and increased bandwidth connecting both DA to two logical loops
These two switched point-to-point loops to each drive, plus connecting both DAs to each
switch, accounts for the following:
There is no arbitration competition and interference between one drive and all the other
drives because there is no hardware in common for all the drives in the FC-AL loop. This
leads to an increased bandwidth utilizing the full speed of a Fibre Channel for each
individual drive. Note that the external transfer rate of a Fibre Channel DDM is 200 MBps.
Doubles the bandwidth over conventional FC-AL implementations due to two
simultaneous operations from each DA to allow for two concurrent read operations and
two concurrent write operations at the same time.
In addition to the superior performance, we must not forget the improved RAS over
conventional FC-AL. The failure of a drive is detected and reported by the switch. The
switch ports distinguish between intermittent failures and permanent failures. The ports
understand intermittent failures which are recoverable and collect data for predictive failure
statistics. If one of the switches itself fails, a disk enclosure service processor detects the
failing switch and reports the failure using the other loop. All drives can still connect
through the remaining switch.
This discussion has just outlined the physical structure. A virtualization approach built on top
of the high performance architecture design contributes even further to enhanced
performance. See Chapter 6, “Virtualization concepts” on page 91.
To host servers
Adapter Adapter
DA
Adapter Adapter
Fibre Channel Fibre Channel
Protocol Proc Protocol Proc
The RAID device adapter is built on PowerPC technology with four 2 Gbps Fibre Channel
ports and high function, high performance ASICs. Each port provides up to five times the
throughput of the SSA-based DA ports used in predecessor disk subsystems. Each single
Fibre Channel protocol processor satisfies both Fibre Channel 2 Gbps ports at full speed.
Note that each DA performs the RAID logic and frees up the processors from this task. The
actual throughput and performance of a DA is not only determined by the 2 Gbps ports and
hardware used, but also by the firmware efficiency.
Each port continues the tradition of providing industry-leading throughput and I/O rates for
FICON and FCP.
To host servers
Adapter Adapter
Storage server
PowerPC
HA
Adapter Adapter
With FC adapters that are configured for FICON, the DS8000 series provides the following
configuration capabilities:
Either fabric or point-to-point topologies
A maximum of 64 host adapter ports on the DS8100 Model 931, and a maximum of 128
host adapter ports on DS8300 Model 932 and Model 9B2
A maximum of 509 logins per fibre-channel port
A maximum of 8,192 logins per storage unit
A maximum of 1280 logical paths on each fibre-channel port
Access to all control-unit images over each FICON port
A maximum of 512 logical paths per control unit image.
FICON host channels limit the number of devices per channel to 16,384. To fully access
65,280 devices on a storage unit, it is necessary to connect a minimum of four FICON host
channels to the storage unit. This way using a switched configuration, you can expose 64
control-unit images (16,384 devices) to each host channel.
The front end with the 2 Gbps or 4 Gbps ports scales up to 128 ports for a DS8300. This
results in a theoretical aggregated host I/O bandwidth of 128 times 2 Gbps for the 2Gbps
ports and outperforms an ESS by a factor of eight. The DS8100 still provides four times more
bandwidth at the front end than an ESS.
The following sections discuss configuration and performance aspects based on the two-way
processor complexes used in the DS8100.
Among the most exciting capabilities the System p inherited from System z are the dynamic
LPAR mode and the micro partitioning capability. This System p-based functionality has the
potential to be exploited also in future disk storage server enhancements. For details on LPAR
implementation in the DS8000 see Chapter 3, “Storage system LPARs (logical partitions)” on
page 27.
Besides the self-healing features and advanced RAS attributes, the RIO-G structures provide
a very high I/O bandwidth interconnect with DAs and HAs to provide system-wide balanced
aggregated throughput from top to bottom. A simplified view is in Figure 10-5.
To host servers
Storage server
Processor Memory
Memory
Figure 10-5 Standard System p5 2-way SMP processor complexes for DS8100 Model 931
The smallest processor complex within a DS8100 is the POWER5+ two-way SMP processor
complex. The dual-processor complex approach allows for concurrent microcode loads,
transparent I/O failover and failback support, and redundant, hot-swapable components.
HA
Fibre Channel Fibre Channel
Protocol Proc Protocol Proc
PowerPC
RIO-G Module
Server 0 Server 1
L1,2 L1,2
Memory Processor
Memory Processor Memory
Memory
L3
Memory
L1,2
Memory Processor RIO-G Interconnect Processor
L1,2
Memory
L3
Memory
DA
I/O enclosure Fibre Channel
Protocol Proc
Fibre Channel
Protocol Proc
DA
4 HAs
Figure 10-6 DS8100-931 with four I/O enclosures
Figure 10-6 provides a less abstract view and outlines some details on the dual 2-way
processor complex of a DS8100 Model 931, its gates to host servers through HAs, and its
connections to the disk storage back end through the DAs.
Each of the two processor complexes is interconnected through the System p-based RIO-G
interconnect technology and includes up to four I/O enclosures which equally communicate to
either processor complex. Note that there is some affinity between the disk subsystem and its
individual ranks to either the left processor complex, server 0, or to the right processor
complex, server 1. This affinity is established at creation of an extent pool.
Each single I/O enclosure itself contains six Fibre Channel adapters:
Two DAs which install in pairs
Four HAs which install as required
Although each HA can communicate with each server, there is some potential to optimize
traffic on the RIO-G interconnect structure. RIO-G provides a full duplex communication with
1 Gbps in either direction. There is no such thing as arbitration. Figure 10-6 shows that the
two left-most I/O enclosures might communicate with server 0, each in full duplex. The two
right-most I/O enclosures communicate with server 1, also in full duplex mode. This results in
a potential of 8 GBps in total just for this single structure. Basically there is no affinity between
HA and server. As we see later, the server which owns certain volumes through its DA,
communicates with its respective HA when connecting to the host.
Server 0 Server 1
L1,2 L1,2
Memory Processor
Memory Processor Memory
Memory
L3
Memory
L1,2
Memory Processor
RIO-G Interconnect Processor
L1,2
Memory
L3
Memory
20 port switch
ooo
16 DDM
20 port switch
20 port switch
ooo
16 DDM
20 port switch
Figure 10-7 Fibre Channel switched backend connect to processor complexes - partial view
All I/O enclosures within the RIO interconnect fabric are equally served from either processor
complex.
Each I/O enclosure contains two DAs. Each DA with its four ports connects to four switches to
reach out to two sets of 16 drives or disk drive modules (DDMs) each. Note that each 20-port
switch has two ports to connect to the next switch pair with 16 DDMs when vertically growing
within a DS8000. As outlined before, this dual two logical loop approach allows for multiple
concurrent I/O operations to individual DDMs or sets of DDMs and minimizes arbitration
through the DDM/switch port mini loop communication.
Server 0 Server 1
I/O enclosure I/O enclosure
L1,2 L1,2
Memory Processor
Memory Processor Memory
Memory
L3
Memory
L1,2
Memory Processor RIO-2 Interconnect Processor
L1,2
Memory
L3
L3
Memory
Memory
Server 0 Server 1
L1,2 L1,2
Memory Processor
Memory Processor Memory
Memory
L3
Memory
L1,2
Memory Processor RIO-2 Interconnect Processor
L1,2
Memory
L3
L3
Memory
Memory
L3
Memory
L1,2
Memory Processor RIO-2 Interconnect Processor
L1,2
Memory
L3
L3
Memory
Memory
RIO-G Module
RIO-G Module
POWER5 4-way SMP POWER5 4-way SMP
Figure 10-8 DS8100 to DS8300 scale performance linearly - view without disk subsystems
Although Figure 10-8 does not display the back-end part, it can be derived from the number
of I/O enclosures, which suggests that the disk subsystem also doubles, as does everything
else, when switching from a DS8100 to an DS8300. Doubling the number of processors and
I/O enclosures accounts also for doubling the performance or even more.
Again note here that a virtualization layer on top of this physical layout contributes to
additional performance potential.
the DS8000 to cooperate with the host systems in manners beneficial to the overall
performance of the systems.
10.2.1 End to End I/O Priority — synergy with System p AIX and DB2
End to end I/O priority is a new addition, requested by IBM, to SCSI T10 standard. This
feature allows trusted applications to override the priority given to each I/O by the operating
system. This is only applicable for raw volumes (no file system) and with the 64 bit kernel.
Currently, AIX supports this feature in conjunction with DB2. The priority is delivered to
storage subsystem in FCP Transport Header.
The priority of an AIX process can be 0 (no assigned priority) or any integer value from 1
(highest priority) to 15 (lowest priority). All I/O requests associated with a given process
inherit its priority value, but with end to end priority, DB2 can change this value for critical data
transfers. At the DS8000, the host adapter will give preferential treatment to higher priority
I/O, improving performance for specific requests to deemed important by the application such
as ones that might be prerequisites for other (eg - DB2 logs).
With the implementation of Cooperative Caching the AIX operating system allows trusted
applications, such as DB2, to provide cache hints to the DS8000. This improves the
performance of the subsystem by keeping more of the repeatedly accessed data within the
cache. Cooperative caching is supported in System p AIX with Multipath I/O (MPIO) Path
Control Module (PCM) provided with Subsystem Device Driver (SDD). It is only applicable for
raw volumes (no file system) and with the 64 bit kernel.
10.2.3 Long Busy Wait Host Tolerance — synergy with System p AIX
Another new addition to SCSI T10 standard is SCSI long busy wait, which provides the target
system a method by which to specify not only that it is busy but also how long the initiator
should wait before retrying an I/O.
This information, provided in the Fiber Channel Protocol (FCP) status response, prevents the
initiator from retrying too soon only to fail again. This in turn reduces unnecessary requests
and potential I/O failures due to exceeding a set threshold for the number of retries. IBM
System p AIX supports for SCSI long busy wait with MPIO and it is also supported by the
DS8000.
The factors that have to be considered to determine the proper cache size are:
The total amount of disk capacity that the DS8000 will hold
The characteristic access density (I/Os per GB) for the stored data
The characteristics of the I/O workload (cache friendly, unfriendly, standard; block size;
random or sequential; read/write ratio; I/O rate)
If you do not have detailed information regarding the access density and the I/O operations
characteristics, but you only know the usable capacity, you can estimate between 2 GB and
4 GB for the size of the cache per 1 TB of storage, as a general rule of thumb.
As is common for data placement and to optimize the DS8000 resources utilization, you
should:
Equally spread the LUNs across the DS8000 servers. Spreading the LUNs equally on
rank group 0 and 1 will balance the load across the DS8000 units.
Use as many disks as possible.
Distribute across DA pairs and RIO-G loops.
Stripe your logical volume across several ranks.
Consider placing specific database objects (such as logs) on different ranks.
All disks in the storage disk subsystem should have roughly the equivalent utilization. Any
disk that is used more than the other disks will become a bottleneck to performance. A
practical method is to make extensive use of volume level striping across disk drives.
Drive size and architecture dictate the amount of data per drive head. As the physical drive
size increases, so does the potential workload for arrays and logical volumes on those drives.
Keep this in mind when planning for solutions requiring high IO rates and fast response times.
One way to counter this is with faster drives speeds, but for the best performance, use arrays
of small high speed drives (e.g. - 15k rpm, 73 GB). The higher rotational speed reduces seek
time, thus improving performance.
To estimate the number and capacity of disk drive sets needed to fulfill your storage capacity
requirements, the Capacity Magic tool can be used. It is a friendly tool and easy to use, that
will also help you determine the requirements for any growth in capacity that you may want to
estimate.
DS8000 logical volumes are composed of extents. An extent pool is a logical construct to
manage a set of extents. One or more ranks with the same attributes can be assigned to an
extent pool. One rank can be assigned to only one extent pool. To create the logical volume,
extents from one extent pool are concatenated. If an extent pool is made up of several ranks,
a LUN can potentially have extents on different ranks and so be spread over those ranks.
Note: We recommend assigning one rank per extent pool to control the placement of the
data. When creating a logical volume in an extent pool made up of several ranks, the
extents for this logical volume are taken from the same rank if possible.
However, to be able to create very large logical volumes, you must consider having extent
pools that span more than one rank. In this case, you will not control the position of the LUNs
and this may lead to an unbalanced implementation, as shown in Figure 10-9 —right side.
Extent
Rank 2
Extent pool 2
1GB 2GB LUN 2 Rank 6
Extent
1GB
Extent pool 3
2GB LUN 3 Rank 7
Rank 3
Extent pool 4 Rank 8
2GB LUN 4
Rank 4
Combining extent pools made up of one rank and then LVM striping over LUNs created on
each extent pool, will offer a balanced method to evenly spread data across the DS8000 as
shown in Figure 10-9 —left side.
Note: The recommendation is to use host striping wherever possible to distribute the
access patterns across the physical resources of the DS8000.
The stripe size has to be large enough to keep sequential data relatively close together, but
not too large so as to keep the data located on a single array.
The recommended stripe sizes that should be defined using your host’s logical volume
manager are in the range of 4 MB to 64 MB. You should choose a stripe size close to 4 MB if
you have a large number of applications sharing the arrays and a larger size when you have
very few servers or applications sharing the arrays.
10.3.6 Determining the number of connections between the host and DS8000
When you have determined your workload requirements in terms of throughput, you have to
choose the appropriate number of connections to put between your open systems servers
and the DS8000 to sustain this throughput.
A Fibre Channel host port can sustain a data transfer of 206 MB/s. As a general
recommendation, you should at least have two FC connections between your hosts and your
DS8000.
The dynamic I/O load-balancing option (default) of SDD is recommended to ensure better
performance because:
SDD automatically adjusts data routing for optimum performance. Multipath load
balancing of data flow prevents a single path from becoming overloaded, causing
input/output congestion that occurs when many I/O operations are directed to common
devices along the same input/output path.
The path to use for an I/O operation is chosen by estimating the load on each adapter to
which each path is attached. The load is a function of the number of I/O operations
currently in process. If multiple paths have the same load, a path is chosen at random from
those paths.
For more information on the SDD see Section 15.1.4, “Multipathing support — Subsystem
Device Driver (SDD)” on page 290
The DS8000 host adapters have no server affinity, but the device adapters and the rank have
server affinity. Illustrated in Figure 10-10 on page 180 is a host that is connected through two
FC adapters to two DS8000 host adapters located in different I/O enclosures. The host has
access to LUN1, which is created in the extent pool 1 controlled by the DS8000 server 0. The
host system sends read commands to the storage server. When a read command is
executed, one or more logical blocks are transferred from the selected logical drive through a
host adapter over an I/O interface to a host. In this case the logical device is managed by
server 0 and the data is handled by server 0. The read data to be transferred to the host must
first be present in server 0's cache. When the data is in the cache it is then transferred
through the host adapters to the host.
IBM
Reads
Reads
L1,2
Memory L1,2
Memory Processor
Processor Memory
Memory
SERVER 0
SERVER 1
L3
L1,2
Memory
RIO-2 Interconnect L1,2
Memory Processor L3
Processor Memory
Memory
20 port switch
DA 20 port switch DA
DAs with an affinity to server 0 LUN1 ooo DAs with an affinity to server 1
16 DDM
20 port switch
Each I/O enclosure can hold up to four HAs. The example in Figure 10-11 shows only eight
FICON channels connected to the first two I/O enclosures. Not shown is a second FICON
director, which connects in the same fashion to the remaining two I/O enclosures to provide a
total of 16 FICON channels in this particular example. The DS8100 disk storage subsystem
provides up to 64 FICON channel ports. Again note the very efficient FICON implementation
in the DS8000 FICON ports.
FICON
HA
Director
Server 0 Server 1
L1,2 L1,2
Memory Processor
Memory Processor Memory
Memory
L3
L1,2
Memory Processor
RIO-G Interconnect L1,2
Memory
L3
Memory Processor Memory
A sizing approach to follow would be to propose how many ESS 800s might be consolidated
into a DS8000 series model. From there you can derive the number of ESS 750s, ESS F20s,
and ESS E20s which can collapse into a DS8000. The older ESS models have a known
relationship to the ESS 800. Further considerations are, for example, the connection
technology used, like ESCON, FICON, or FICON Express channels, and the number of
channels.
Generally speaking, a properly configured DS8100 has the potential to provide the same or
better numbers than two ESS 800s. Since the ESS 800 has the performance capabilities of
two ESS F20s, a properly configured DS8100 can replace four ESS F20s. As the DS8000
series scales linearly, a well configured DS8300 has the potential to have the same or better
numbers concerning I/O rates, sequential bandwidth, and response time than two DS8100 or
four ESS 800s. Since the ESS 800 has roughly the performance potential of two ESS F20s, a
corresponding number of ESS F20s can be consolidated. This applies also to the ESS 750,
which has a similar performance behavior to that of an ESS F20.
When comparing the DS8000 series Turbo models 931, 932, and 9B2 against the
predecessor DS8000 series models 921, 922, and 9A2, consider that the Turbo models
feature the IBM POWER5+ processor which can enable up to a 15% performance
improvement in I/O operations per second in transaction processing workload environments
compared to the IBM POWER5 processor of the 921/922/9A2 models.
Processor memory or cache can grow to up to 256 GB in the DS8300 and to 128 GB for the
DS8100. This processor memory is subdivided into a data in cache portion, which holds data
in volatile memory, and a persistent part of the memory, which functions as NVS to hold
DASD fast write (DFW) data until de-staged to disk.
It is not just the pure cache size which accounts for good performance figures. Economical
use of cache, like 4 KB cache segments and smart, adaptive caching algorithms, are just as
important to guarantee outstanding performance. This is implemented in the DS8000 series.
Besides the potential for sharing data on a DS8000 processor memory level, cache is the
main contributor to good I/O performance.
Use your current configuration and workload data and consider these guidelines:
Choose a cache size for the DS8000 series which has a similar ratio between cache size
and disk storage capacity to that of the currently used configuration.
When you consolidate multiple disk storage units, configure the DS8000 processor
memory or cache size as the sum of all cache from your current disk storage units.
For example, consider a DS8100 replacing four ESS F20s with 3.2 TB and 16 GB cache
each. The ratio between cache size and disk storage for the ESS F20s is 0.5% with 16
GB/3.2 TB. The new DS8100 is configured with 18 TB to consolidate 4 x 3.2 TB plus some
extra capacity for growth. This would require 90 GB of cache to keep the original
cache-to-disk storage ratio. Round up to the next available memory size, which is 128 GB for
this DS8100 configuration.
This ratio of 0.5% cache to backstore is considered high performance for z/OS environments.
Standard performance suggests a ratio of 0.2% cache to backstore ratio.
FICON ports. So four ESS 800s with eight FICON channels each connected to IBM 9672
G5 or G6 servers might end up in a single DS8300 with 32 FICON channels.
When migrating not only to DS8000 models, but also from 1 Gbps FICON to FICON
Express channels at 2 Gbps, you can consider consolidating the number of channels to
about 2/3 of the original number of channels. Use at least four FICON channels per
DS8000 —by the way, when we write about FICON channels, we mean FICON ports in
the disk storage units.
Coming from FICON Express channels, you should keep a minimum of four FICON ports.
You might consider using 25% fewer FICON ports in the DS8000 than the aggregated
number of FICON 2 Gbps ports from the source environment. For example, when you
consolidate two ESS 800s with eight FICON 2 Gbps ports each to a DS8100, plan for a
minimum of 12 FICON ports on the DS8100.
Another example, with four ESS F20s each with eight FICON channels, might collapse into
about 20 FICON ports when changing to a connectivity speed of 2 Gbps for the target
DS8100 or DS8300.
You can approach this from the disk side and look at some basic disk figures. Fibre Channel
disks, for example, at 10k RPM provide an average seek time of approximately 5 ms and an
average latency of 3 ms. For transferring only a small block, the transfer time can be
neglected. This is an average 8 ms per random disk I/O operation or 125 I/Os per second. A
15k RPM disk provides about 200 random I/Os per second for small block I/Os. A combined
number of 8 disks is then good for 1,600 I/Os per second when they spin at 15k per minute.
Reduce the number by 12.5% when you assume a spare drive in the 8 pack. Assume further
a RAID-5 logic over the 8 packs.
Back at the host side, consider an example with 4,000 I/Os per second, a read to write ratio of
3 to 1, and 50% read cache hits. This leads to the following I/O numbers:
3,000 read I/Os per second.
1,500 read I/Os must read from disk.
1,000 writes with RAID-5 and assuming the worst case results in 4,000 disk I/Os.
This totals to 4,500 disk I/Os.
With 15K rpm DDMs you need the equivalent of three 8 packs to satisfy the I/O load from the
host for this example. Depending on the required capacity, you then decide the disk capacity,
provided each desired disk capacity has 15k rpm. When the access density is less and you
need more capacity, follow the example with higher capacity disks, which usually spin at a
slower speed like 10k rpm.
For a single disk drive, various disk vendors provide on their internet product sites the disk
specifications like for example:
146 GB DDM with 10K rpm deliver a sustained transfer rate between 38 and 68 MB/s or
53 MB/s on average.
73 GB DDM with 15K rpm transfers between 50 and 75 MB/s or 62.5 MB/s on average.
The 73 GB DDMs have about 18% more sequential capability than the 146 GB DDM, but
60% more random I/O potential —the I/O characteristic is another aspect to consider when
deciding the disk and disk array size. While this discussion is theoretical in approach, it
provides a first impression.
Once the speed of the disk has been decided, the capacity can be calculated based on you
storage capacity needs and the effective capacity of the RAID configuration you will use. For
this you can use Table 8-9 on page 143.
To estimate the number and capacity of disk drive sets needed to fulfill your storage capacity
requirements, the Capacity Magic tool can be used. It is a friendly tool and easy to use, that
will also help you determine the requirements for any growth in capacity that you may want to
estimate.
Note the independence of LSSs from ranks in the DS8000. Because an LSS is congruent
with a System z LCU, we need to understand the implications. It is now possible to have
volumes within the very same LCU, which is the very same LSS, but these volumes might
reside in different ranks.
A horizontal pooling approach assumes that volumes within a logical pool of volumes, like all
DB2 volumes, are evenly spread across all ranks. This is independent of how these volumes
are represented in LCUs. The following sections assume horizontal volume pooling across
ranks, which might be congruent with LCUs when mapping ranks accordingly to LSSs.
Due to the virtualization of the Fibre Channel switched disk subsystem you might consider
creating as many extent pools as there are RAID ranks in the DS8000. This would then work
similar to what is currently in the ESS. With this approach you can control the placement of
each single volume and where it ends up in the disk subsystem.
In the example in Figure 10-12 on page 185, each rank is in its own extent pool. The evenly
numbered extent pools have an affinity to the left server, server 0. The odd number extent
pools have an affinity to the right server, server 1. When a rank is subdivided into extents it
gets assigned to its own extent pool.
HA
L1,2
Memory L1,2
Memory Processor
Processor Memory Memory
L1,2 L1,2
L3 L3
Memory
Memory Processor Processor Memory
Memory
Server0 Server1
RIO-G Module RIO-G Module
POWER5 2-way SMP POWER5 2-way SMP
DAs
Rank Extent pool0 Extent pool2 Extent pool4 Extent pool1 Extent pool3 Extent pool5
1 3 5 2 4 6
7 9 11 8 10 12
Extent pool6 Extent pool8 Extent pool10 Extent pool7 Extent pool9 Extent pool11
Figure 10-12 Extent pool affinity to processor complex with one extent pool for each rank
Now all volumes which are comprised of extents out of an extent pool have also a respective
server affinity when scheduling I/Os to these volumes.
This allows you to place certain volumes in specific ranks to avoid potential clustering of many
high activity volumes within the same rank. You can create SMS storage groups which are
congruent to these extent pools to ease the management effort of such a configuration. But
you can still assign multiple storage groups when you are not concerned about the placement
of less active volumes.
Figure 10-12 also indicates that there is no affinity nor a certain preference between HA and
processor complexes or servers in the DS8000. In this example either one of the two HAs can
address any volume in any of the ranks, which range here from rank number 1 to 12. Note
there is an affinity of DAs to the processor complex. A DA pair connects to two switches
respectively to two pairs of switches —see Figure 10-2 on page 169. The first DA of this DA
pair connects to the left processor complex or server 0. The second DA of this DA pair
connects to the other processor complex or server 1.
HA
L1,2
Memory L1,2
Memory Processor
Processor Memory Memory
L1,2 L1,2
L3 L3
Memory
Memory Processor Processor Memory
Memory
Server0 Server1
RIO-G Module RIO-G Module
POWER5 2-way SMP
POWER5 2-way SMP
DAs
Rank 1 3 2 4
5 7 6 8
9 11 10 12
Again what is obvious here is the affinity between all volumes residing in extent pool 0 to the
left processor complex, server 0, and vice versa for the volumes residing in extent pool 1 and
their affinity to the right processor complex or server 1.
When creating volumes there is no straightforward approach to place certain volumes into
certain ranks. For example, when you create the first 20 DB2 logging volumes, they would be
allocated in a consecutive fashion in the first rank. The concerned RAID site would then host
all these 20 logging volumes. Now with Array Across Loops (AAL) and the high performance
and large bandwidth capabilities of the Fibre Channel switched disk storage subsystem, this
might not be an issue. You may choose to control the placement of your most performance
critical volumes. This might lead to a compromise between both approaches.
Create two general extent pools for all the average workload and the majority of the volumes
and subdivide these pools evenly between both processor complexes or servers. These pools
contain the majority of the installed ranks in the DS8000. Then you might consider two or four
smaller extent pools with dedicated ranks for high performance workloads and their
respective volumes. You may consider defining storage groups accordingly which are
congruent to the smaller extent pools.
Server0 Server1
DAs
POWER5 2-way POWER5 2-way
SMP SMP
SGLOG1 SGLOG2
Rank Extent pool0 Extent pool2 Extent pool3 Extent pool1
1 1
3 5 5 3
7 9 SGPRIM 9 7
SGHPC1 11 11 SGHPC2
Consider grouping the two larger extent pools into a single SMS storage group. SMS will
eventually spread the workload evenly across both extent pools. This allows a system
managed approach to place data sets automatically in the right extent pools.
With more than one DS8000 you might consider configuring each DS8000 in a uniform
fashion. We recommend grouping all volumes from all the large extent pools into one large
SMS storage group, SGPRIM. Cover the smaller, high performance extent pools through
discrete SMS storage groups for each DS8000.
With two of the configurations displayed in Figure 10-14, this ends up with one storage group,
SGPRIM, and six smaller storage groups. SGLOG1 contains Extent pool0 in the first DS8100
and the same extent pool in the second DS8100. Similar considerations are true for
SGLOG2. For example, in a dual logging database environment this allows you to assign
SGLOG1 to the first logging volume and SGLOG2 for the second logging volume. For very
demanding I/O rates and to satisfy a small set of volumes, you might consider keeping Extent
pool 4 and Extent pool 5 in both DS8100s separate, through four distinct storage groups,
SGHPC1-4.
Figure 10-14 shows, again, that there is no affinity between HA and processor complex or
server. Each I/O enclosure connects to either processor complex. But there is an affinity
between extent pool and processor complex and, therefore, an affinity between volumes and
processor complex. This requires some attention, as outlined previously, when you define
your volumes.
systems, helping the System z servers running applications to concurrently share the same
logical volumes.
The ability to do multiple I/O requests to the same volume nearly eliminates IOSQ time, one
of the major components in z/OS response time. Traditionally, access to highly active volumes
has involved manual tuning, splitting data across multiple volumes, and more. With PAV and
the Workload Manager, you can almost forget about manual performance tuning. WLM
manages PAVs across all the members of a sysplex too. This way, the DS8000 in conjunction
with z/OS has the ability to meet the performance requirements by its own.
Figure 10-15 illustrates the traditional z/OS behavior without PAV, where subsequent
simultaneous I/Os to volume 100 are queued while volume 100 is still busy with a preceding
I/O.
One I/O to
one volume
System z at one time System z
100
Figure 10-15 Traditional z/OS behavior
From a performance standpoint, it did not make sense to send more than one I/O at a time to
the storage disk subsystem, because the hardware could process only one I/O at a time.
Knowing this, the z/OS systems did not try to issue another I/O to a volume —in z/OS
represented by a Unit Control Block (UCB)— while an I/O was already active for that volume,
as indicated by a UCB busy flag; see Figure 10-15. Not only were the z/OS systems limited to
processing only one I/O at a time, but also the storage subsystems accepted only one I/O at a
time from different system images to a shared volume, for the same reasons mentioned
above.
z/OS
Single image
System z
Physical layer
The DS8000 has the capability to do more than one I/O to a CKD volume. Using alias address
in addition to the conventional base address, a z/OS host can use several UCBs for the same
logical volume instead of one UCB per logical volume. For example, base address 100 may
have alias addresses 1FF and 1FE which allows for three parallel I/O operations to the same
volume; see Figure 10-16.
This feature that allows parallel I/Os to a volume from one host is called Parallel Access
Volume (PAV).
The two concepts that are basic in the PAV functionality are:
Base address: the base device address is the conventional unit address of a logical
volume. There is only one base address associated with any volume.
Alias address: an alias device address is mapped to a base address. I/O operations to an
alias run against the associated base address storage space. There is no physical space
associated with an alias address. You can define more than one alias per base.
Alias address have to be defined to the DS8000 and to the IODF file. This association is
predefined and adding new aliases can be done non-disruptively. Still, the association
between base and alias is not fixed —alias address can be assigned to different base
address by the z/OS Workload Manager.
For guidelines on PAV definition and support, see 16.3.2, “Parallel Access Volumes (PAV)
definition” on page 338.
z/OS recognizes the aliases that are initially assigned to a base during the NIP (Nucleus
Initialization Program) phase. If dynamic PAVs are enabled, the WLM can reassign an alias to
another base by instructing the IOS to do so when necessary; seeFigure 10-17.
WLM
IOS
Assign to base 100
z/OS’s Workload Manager in Goal mode tracks the system workload and checks if the
workloads are meeting their goals established by the installation; see Figure 10-18.
WLM also keeps track of the devices utilized by the different workloads, accumulates this
information over time, and broadcasts it to the other systems in the same sysplex. If WLM
determines that any workload is not meeting its goal due to IOSQ time, WLM will attempt to
find an alias device that can be reallocated to help this workload achieve its goal; see
Figure 10-18.
As mentioned before, the Workload Manager (WLM) must be in Goal mode to cause PAVs to
be shifted from one logical device to another.
The movement of an alias from one base to another is serialized within the sysplex. IOS
tracks a token for each PAV enabled device. This token is updated each time an alias change
is made for a device. IOS and WLM exchange the token information. When the WLM instructs
IOS to move an alias, WLM also presents the token. When IOS has started a move and
updated the token, all affected systems are notified of the change through an interrupt.
DSK001
Figure 10-19 z/VM support of PAV volumes dedicated to a single guest virtual machine
DSK001
In this way, PAV provides to the z/VM environments the benefits of a greater I/O performance
(throughput) by reducing I/O queueing.
With the SPE introduced with z/VM 5.2.0 and APAR VM63952 additional enhancements are
available when using PAV with z/VM. For more information see Section 16.4, “z/VM
considerations” on page 340.
Appl. A Appl. B
UCB 100 UCB 100
Multiple
System z Allegiance System z
Physical
layer
Figure 10-21 Parallel I/O capability with Multiple Allegiance
The DS8000 accepts multiple I/O requests from different hosts to the same device address,
increasing parallelism and reducing channel overhead. In older storage disk subsystems, a
device had an implicit allegiance, that is, a relationship created in the control unit between the
device and a channel path group when an I/O operation is accepted by the device. The
allegiance causes the control unit to guarantee access (no busy status presented) to the
device for the remainder of the channel program over the set of paths associated with the
allegiance.
With Multiple Allegiance, the requests are accepted by the DS8000 and all requests will are
processed in parallel, unless there is a conflict when writing data to a particular extent of the
CDK logical volume; see Figure 10-21 on page 192.
Still, good application software access patterns can improve the global parallelism by
avoiding reserves, limiting the extent scope to a minimum, and setting an appropriate file
mask, for example, if no write is intended.
In systems without Multiple Allegiance, all except the first I/O request to a shared volume
were rejected, and the I/Os were queued in the System z channel subsystem, showing up as
PEND time in the RMF reports.
10.4.11 HyperPAV
The DS8000 series offers enhancements to Parallel Access Volumes (PAV) with support for
HyperPAV, which is designed to enable applications to achieve equal or better performance
than PAV alone, while also using the same or fewer operating system resources.
Multiple Allegiance and PAV allow multiple I/Os to be executed concurrently against the same
volume:
With Multiple Allegiance, the I/Os are coming from different system images
With PAV, the I/Os are coming from the same system image:
– Static PAV: aliases are always associated with the same base addresses
– Dynamic PAV: aliases are assigned up front but can be reassigned to any base
address as need dictates by means of the Dynamic Alias Assignment function of the
Workload Manager—reactive alias assignment.
HyperPAV
With HyperPAV an on demand proactive assignment of aliases is now possible, see
Figure 10-23.
HyperPAV allows an alias address to be used to access any base on the same control unit
image per I/O base. This capability also allows different HyperPAV hosts to use one alias to
access different bases which reduces the number of alias addresses required to support a set
of bases in a System z environment with no latency in targeting an alias to a base. This
functionality is also designed to enable applications to achieve equal or better performance
than possible with the original PAV feature alone while also using the same or fewer operating
system resources.
Benefits of HyperPAV
HyperPAV has been designed to:
Provide an even more efficient Parallel Access Volumes (PAV) function
Help customers who implement larger volumes to scale I/O rates without the need for
additional PAV alias definitions
Exploit FICON architecture to reduce overhead, improve addressing efficiencies, and
provide storage capacity and performance improvements:
– More dynamic assignment of PAV-aliases improves efficiency
– Number of PAV-aliases needed may be reduced, taking fewer from the 64K device
limitation and leaving more storage for capacity use
Enable a more dynamic response to changing workloads
Simplified management of aliases
Enable users to stave off migration to larger volume sizes.
For support information see sections 16.3.3, “HyperPAV — z/OS support” on page 339 and
16.4.3, “PAV and HyperPAV — z/VM support” on page 340.
Priority queueing
I/Os from different z/OS system images can be queued in a priority order. It is the z/OS
Workload Manager that makes use of this priority to privilege I/Os from one system against
the others. You can activate I/O Priority Queuing in WLM Service Definition settings. WLM
has to run in Goal mode.
System A System B
WLM WLM
IO Queue
for I/Os that IO with
have to be Priority
queued X'21'
When a channel program with a higher priority comes in and is put in front of the queue of
channel programs with lower priority, the priority of the low-priority programs is also
increased; see Figure 10-24. This prevents high-priority channel programs from dominating
lower priority ones and gives each system a fair share.
11
For the planning activities related to the activation of the licensed functions see also the
discussion in 9.4.5, “Activation of Advanced Function licenses” on page 155.
The 242x/2107 licensed function indicator feature numbers enable the technical activation of
the function subject to the client applying a feature activation code made available by IBM.
The 239x/2244 licensed function authorization feature numbers establish the extent of
authorization for that function on the 242x or 2107 machine for which it was acquired.
Table 11-1 DS8000 series Turbo Models 93x/9B2 with Enterprise Choice length of warranty
Licensed function for IBM 242x indicator IBM 239x function authorization
Turbo Models 93x/9B2 —with feature numbers model and feature numbers
Enterprise Choice warranty
Metro Mirror Add on 0754 and 75xx 239x Model LFA, 75xx
Global Mirror Add on 0756 and 75xx 239x Model LFA, 75xx
Remote Mirror for z/OS 0760 and 76xx 239x Model LFA, 76xx
Parallel access volumes 0780 and 78xx 239x Model LFA, 78xx
Table 11-2 DS8000 series Turbo Models 93x/9B2 without Enterprise Choice length of warranty
Licensed function for IBM 2107 indicator IBM 2244 function authorization
Turbo Models 93x/9B2 feature number model and feature numbers
—without Enterprise Choice w.
Note: For a detailed explanation of the features involved and the considerations you must
have when ordering DS8000 licensed functions refer to the announcement letters:
IBM System Storage DS8000 Series (IBM 2107 and IBM 242x)
IBM System Storage DS8000 — Function Authorizations (IBM 2244 or IBM 239x).
You may activate the license keys all at the same time (for example, on initial activation of the
storage unit) or activate them individually (for example, additional ordered keys).
Before connecting to the IBM DSFA site to obtain your feature activation codes, ensure that
you have the following items:
The IBM License Function Authorization documents. If you are activating codes for a new
storage unit, these documents are included in the shipment of the storage unit. If you are
activating codes for an existing storage unit, IBM sends these documents to you in an
envelope.
A diskette or USB memory device for downloading your activation codes into a file if you
cannot access the DS Storage Manager from the system that you are using to access the
DSFA Web site. Instead of downloading the activation codes in softcopy format, you can
also print the activation codes and manually enter them using the DS Storage Manager
GUI. However, this is slow and error prone because the activation keys are 32-character
long strings.
For a discussion of the activities in preparation to the activation of the licensed functions see
also 9.4.5, “Activation of Advanced Function licenses” on page 155.
1. Start the DS Storage Manager application. Log in using an ID with administrator access. If
this is the first time you are accessing the machine, contact your IBM CE for the user ID
and password. After successful login, the DS8000 Storage Manager Welcome panel
3. On the Storage units panel, select the storage unit by clicking the box to the left of it, click
Properties in the Select Action pull-down list. The Storage Unit Properties panel opens
(Figure 11-3).
4. On the Storage Unit Properties panel, click the General tab. Gather the following
information about your storage unit:
– From the MTMS field, note the machine's serial number. The Machine Type - Model
Number - Serial Number (MTMS) is a string that contains the machine type, model
number, and serial number. The last seven characters of the string are the machine's
serial number.
– From the Machine signature field, note the machine signature.
You can use Table 11-4 to document the information. You will later enter the information on
the IBM DSFA Web site.
Machine signature
2. Click the DS8000 series (machine type 2107) link. This will bring you to the Select
DS8000 series machine page (Figure 11-5 on page 203).
Note: The examples we are discussing in this section of the book illustrate the activation of
the licensed functions for 2107-922 and 9A2 models. For this reason the machine type and
function authorizations you see in the screens, correspond to Table 11-3 on page 199. For
the DS8000 series Turbo Models 93x/9B2 the machine types and function authorizations
correspond to Table 11-1 on page 198 and Table 11-2 on page 199.
3. Enter the machine information and click Submit. The View machine summary page opens
(Figure 11-6 on page 204).
The View machine summary page shows the total purchased licenses and how much of
them are currently assigned. The example in Figure 11-6 shows a storage unit where all
licenses have already been assigned. When assigning licenses for the first time, the
Assigned field would show 0.0 TB.
4. Click the Manage activations link. The Manage activations page opens. Figure 11-7 on
page 205 shows the Manage activations page for a 2107 Model 9A2 with two storage
images. For each license type and storage image, enter the license scope (FB, CKD, or
All) and a capacity value (in TB) to be assigned to the storage image. The capacity values
are expressed in decimal terabytes with 0.1 TB increments. The sum of the storage image
capacity values for a license cannot exceed the total license value.
5. When you have entered the values, click Submit. The View activation codes page opens,
showing the license activation codes for the storage images (Figure 11-8 on page 206).
Print the activation codes or click Download to save the activation codes in a file that you
can later import in the DS8000. The file will contain the activation codes for both storage
images.
Note: In most situations, the DSFA application can locate your 2244/239x licensed function
authorization record when you enter the DS8000 (2107 or 242x) serial number and
signature. However, if the 2244/239x licensed function authorization record is not attached
to the 2107/242x record, you must assign it to the 2107/242x record using the Assign
function authorization link on the DSFA application. In this case, you need the 2244/239x
serial number (which you can find on the License Function Authorization document).
Note: The initial enablement of any optional DS8000 licensed function is a concurrent
activity (assuming the appropriate level of microcode is installed on the machine for the
given function). The following activation activities are disruptive and require a machine IML
(Models 921, 931, 922, and 932) or reboot of the affected image (Models 9A2 and 9B2):
Removal of a DS8000 licensed function to deactivate the function.
A lateral change or reduction in the license scope. A lateral change is defined as
changing the license scope from fixed block (FB) to count key data (CKD) or from CKD
to FB. A reduction is defined as changing the license scope from all physical capacity
(ALL) to only FB or only CKD capacity.
Important: Before you begin this task, you must resolve any current DS8000 problems.
Contact IBM support for assistance in resolving these problems.
The easiest way to apply the feature activation codes is to download the activation codes from
the IBM Disk Storage Feature Activation (DSFA) Web site to your local computer and import
the file into the DS Storage Manager. If you can access the DS Storage Manager from the
same computer that you use to access the DSFA Web site, you can copy the activation codes
from the DSFA window and paste them into the DS Storage Manager window. The third
option is to manually enter the activation codes in the DS Storage Manager from a printed
copy of the codes.
1. In the My Work navigation panel on the DS Storage Manager Welcome screen, select, in
order, Real-time manager, Manage hardware, and Storage images. The Storage
images panel opens (Figure 11-9).
2. On the Storage images panel, select a storage image whose activation codes you want to
apply. Select Apply activation codes in the Select Action pull-down list. The Apply
Activation codes panel is displayed (Figure 11-10 on page 208).
If this is the first time you apply activation codes, the fields in the panel are empty;
otherwise, the current license codes and values are displayed in the fields and you can
modify or overwrite them, as appropriate.
3. If you are importing your activation codes from a file that you downloaded from the DSFA
Web site, click Import key file. The Import panel is displayed (Figure 11-12).
Enter the name of your key file, then click OK to complete the import process.
If you did not download your activation codes into a file, manually enter the codes into the
appropriate fields on the Apply Activation codes panel.
4. After you have entered the activation codes, either manually or by importing a key file, click
Apply. The Capacity and Storage type fields will now reflect the license information
contained in the activation codes, as in Figure 11-13.
Note: For the 9A2 and 9B2 models, you need to perform the code activation process for
both storage images, one image at a time.
1. Use the showsi command to display the DS8000 machine signature; see Example 11-1.
2. Obtain your license activation codes from the IBM DSFA Web site. See 11.2.2, “Obtaining
activation codes” on page 202.
3. Use the applykey command to activate the codes and the lskey command to verify which
type of licensed features are activated for your storage unit.
a. Give a applykey command at the dscli command prompt as follows. The -file
parameter specifies the key file. The second parameter specifies the storage image.
dscli> applykey -file c:\2107_7520780.xml IBM.2107-7520781
b. Verify that the keys have been activated for your storage unit by issuing the DS CLI
lskey command as shown in Example 11-2.
Note: For 9A2 and 9B2 models, you need to perform the code activation process for both
storage images, for example, using the serial number from Example 11-2 that would be
IBM.2107-7520781 and IBM.2107-7520782.
For more details on the DS CLI, refer to IBM System Storage DS8000: Command-Line
Interface User´s Guide, SC26-7916.
that are later configured. However, it is possible to return to the DSFA Web site at a later time
and change the scope from CKD or FB to All, or from All to either CKD or FB. In every case, a
new activation code will be generated, which can be downloaded and applied.
When deciding which scope to set, there are several scenarios to consider. Use Table 11-5 to
guide you in your choice. This table applies to both Point-in-Time copy and remote mirror and
copy functions.
6 This function has already been set to All. Leave the scope set to All.
Changing it to CKD or FB at this
point requires a disruptive
outage.
Any scenario that changes from FB or CKD to All will not require an outage. If you choose to
change from All to either CKD or FB, then you will need take a disruptive outage. If you are
absolutely certain that your machine will only ever be used for one storage type (for example,
only CKD or only FB), then you could also quite safely just use the All scope.
Example 11-3 Trying to use a feature for which we are not licensed
dscli> lskey IBM.2107-7520391
Date/Time: 11 November 2005 19:01:44 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520391
Activation Key Capacity (TB) Storage Type
=================================================
Flash Copy 5 FB The scope is currently set to FB
Operating Environment 5 All
Remote Mirror and Copy 5 All
dscli> lsckdvol
Date/Time: 11 November 2005 19:01:52 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520391
Name ID accstate datastate configstate deviceMTM voltype orgbvols extpool cap (cyl)
======================================================================================
- 0000 Online Normal Normal 3390-3 CKD Base - P2 3339
- 0001 Online Normal Normal 3390-3 CKD Base - P2 3339
dscli> lsckdvol
Date/Time: 11 November 2005 19:12:52 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520391
Name ID accstate datastate configstate deviceMTM voltype orgbvols extpool cap (cyl)
======================================================================================
- 0000 Online Normal Normal 3390-3 CKD Base - P2 3339
- 0001 Online Normal Normal 3390-3 CKD Base - P2 3339
dscli> lsckdvol
Date/Time: 11 November 2005 20:34:33 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520391
Name ID accstate datastate configstate deviceMTM voltype orgbvols extpool cap (cyl)
======================================================================================
- 0000 Online Normal Normal 3390-3 CKD Base - P2 3339
- 0001 Online Normal Normal 3390-3 CKD Base - P2 3339
dscli> mkflash 0000:0001 But we are still able to create CKD FlashCopies
Date/Time: 11 November 2005 20:34:42 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7520391
CMUC00137I mkflash: FlashCopy pair 0000:0001 successfully created.
dscli>
In this scenario we have made a downward license feature key change. We must schedule an
outage of the storage image. We should in fact only make the downward license key change
immediately before taking this outage.
Restriction: Making a downward license change and then not immediately performing a
reboot of the storage image is not supported. Do not allow your machine to be in a position
where the applied key is different to the reported key.
At this point this is still a valid configuration. This is because the configured ranks on the
machine total less than 5 TB of storage. In Example 11-7 we then try to create a new rank
that would bring the total rank capacity above 5 TB. This command fails.
To configure the additional ranks, we must first increase the license key capacity of every
installed license. In this example, that would be the FlashCopy license.
To make the calculation we use the lsrank command to determine how many arrays the rank
contains, and whether those ranks are being used for FB or CKD storage. Then we use the
lsarray command to find out the disk size being used by each array. Then we multiply the
disk size (73, 146, or 300) by eight (for eight DDMs in each array site).
In Example 11-8 on page 215, lsrank tells us that rank R0 uses array A0 for CKD storage.
Then lsarray tells us that array A0 uses 300 GB DDMs. So we multiple 300 (the DDM size)
by 8, giving us 300 x 8 = 2400 GB. This means we are using 2400 GB for CKD storage.
Now rank R4 in Example 11-8 is based on array A6. array A6 uses 146 GB DDMs, so we
multiply 146 by 8, giving us 146 x 8 = 1168 GB. This means we are using 1168 GB for FB
storage.
So for CKD scope licenses, we are using 2400 GB. For FB scope licenses, we are using 1168
GB. For licenses with a scope of All, we are using 3568 GB. Using the limits shown in
Example 11-6 on page 214 we are within scope for all licenses.
If we combine Example 11-6, Example 11-7, and Example 11-8, we can also see why the
mkrank command in Example 11-7 failed. In Example 11-7 we tried to create a rank using
array A1. Now array A1 uses 300 GB DDMs. This means that for FB scope and All scope
licenses, we will use 300 x 8 = 2400 GB, more license key. Now from Example 11-6, we had
only 5 TB of FlashCopy license with a scope of All. This means that we cannot have total
configured capacity that exceeds 5000 TB. Since we are already using 3568 GB, the attempt
to use 2400 more GB will fail, since 3568 plus 2400 equals 5968 GB, which is clearly more
than 5000 GB. If we increase the size of the FlashCopy license to 10 TB, then we can have
10000 GB of total configured capacity, so the rank creation will then succeed.
Part 3
Part 3 Storage
Configuration
In this part we discuss the configuration tasks required on your DS8000. The topics covered
include:
Configuration with DS Storage Manager GUI
Configuration with Command Line interface
12
The customization work sheets are very important and need to be completed before the
installation. It is important that this information is entered into the machine so that preventive
maintenance and high availability of the machine is maintained.
The customization work sheets can be found in the IBM System Storage DS8000 Host
Systems Attachment Guide, SC26-7917.
The customization work sheets allow you to specify the initial setup for the following:
Company information. This information allows IBM service personnel to contact you as
quickly as possible to access your storage complex.
Management Console network settings. Allows you to specify the IP address and LAN
settings for your Management Console (MC).
Remote support (includes Call Home and remote service settings). Allows you to specify
whether you want outbound (Call Home) or inbound (remote services) remote support.
Notifications (includes SNMP trap and e-mail notification settings). Allows you to specify
the types of notifications that you want and others may want to receive.
Power control. To select and control the various power modes for the storage complex.
Control Switch settings. Allow you to specify certain DS8000 settings that affect host
connectivity. You are asked to enter these choices on the control switch settings work
sheet so that the service representative can set them during the installation of the
DS8000.
8. Create open systems volumes. Create open systems FB volumes and assign them to one
or more volume groups.
9. Create System z LCUs. Define their type and other attributes such as SSID.
10.Create System z volumes. Create System z CKD base volumes and PAV aliases for them.
The actual configuration can be done using either the DS Storage Manager GUI or DS
Command-Line Interface, or a mixture of both. A novice user may prefer to use the GUI, while
a more experienced user may use the CLI, particularly for some of the more repetitive tasks
such as creating large numbers of volumes.
For a more detailed discussion of how to perform the specific tasks, refer to:
Chapter 11, “Features and license keys” on page 197
Chapter 13, “Configuration with DS Storage Manager GUI” on page 223
Chapter 14, “Configuration with Command Line interface” on page 267
13
In this chapter we discuss the use of the DS Storage Manager for storage configuration of the
DS8000, not for Copy Services configuration. For Copy Services configuration in the DS8000
using the DS SM refer to the following books:
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787
To connect to the DS8000 through the browser, enter the URL of either the default Storage
Hardware Management Console (DS HMC) or the optional DS HMC you may have
purchased. You can connect through either DS HMC, but we recommend that once you start
updating and modifying from one DS HMC, you continue to make your changes through that
DS HMC for the duration of the change. The URL consists of the TCP/IP address, as shown
in Figure 13-1, or a fully qualified name.
Figure 13-1 Entering the URL using the TCP/IP address for the DS HMC
The default user ID is admin and the default password is also admin. The first time you log on
you will be prompted to change the password. Make sure that the new user ID and password
you select for your local GUI installation matches the password of the HMC that you will
connect to.
Note: It is always good practice to create alternate or backup user IDs in case one user ID
gets locked out. At least one of these user IDs and passwords must be the same as the
user ID and password on the HMC that you will connect to. If you intend to connect to
multiple HMCs that have different user IDs and passwords, then you should create a user
ID and password for each of these HMCs.
Exit
Close Task
In the Welcome panel of the DS8000 Storage Manager you find these options:
Show all tasks: opens the Task Manager panel, from where you can end a task or switch
to another task.
Hide Task List: to hide the Task list and expands your work area.
Toggle Banner: to remove the banner with the IBM System Storage Manager logo and
expands the working space.
Information Center: to launch the Information Center. The Information Center is the online
help for the DS8000. The Information Center provides contextual help for each panel, but
also is independently accessible from the Internet.
Close Task: to close the active task.
Exit: to log off from the DS Storage Manager.
On the left side of the panel you have the navigation panel where you have the following
selections:
Real-time manager
Simulated manager
The DS Storage Manager displays the configuration of your DS8000 in tables. To make this
more convenient there are several options you can use.
To download the information from the table click Download. This can be useful if you want
to make a documentation of your configuration. The file is in csv format and can be
opened with a spreadsheet program. This function is also useful if the table on the
DS8000 Manager consists of several pages; the csv file includes all pages.
The Print button opens a new window with the table in HTML format and starts the printer
dialog of your PC if you want to print the table.
The Select Action pull-down menu provides you with specific actions you can perform (for
example, Create).
There are also buttons to set and clear filters so that only specific items are displayed in
the table (for example, only FB ranks will be shown in the table). This can be useful if you
have tables with large numbers of items.
When configuring the storage with the Real-time manager the configuration actions are
performed as soon as the commands are issued. When configuring the storage with the
Simulated manager, the configuration commands will only be applied after exporting the
resultant configuration to an unconfigured DS8000 storage unit.
Attention: Changes to an installed configuration may only be made using the Real-time
manager.
The first step is the creation of the storage complex along with the definition of the hardware
of the storage unit. The storage unit may have one or more storage images.
Many of the logical configuration tasks can be performed independently, for example, you can
create host definitions before you create any storage definitions. However, some tasks do
have dependencies on previous items. This might determine the order in which you perform
the logical configuration. For example, you cannot create any host attachment definitions until
you have configured the I/O ports in the Storage Unit panel.
Many of the task wizards will take you into other wizards to create interdependent items, for
example, if you create a storage unit, but have no storage complex defined, the Create
Storage Unit wizard will take you into the Create Storage Complex wizard. This may seem
confusing, but as long as you have a comprehensive document of your proposed logical
configuration, you can follow the process through the wizards.
Figure 13-4 shows the successful completion of the step to import a storage complex.
The Add Storage Complex page is displayed. Specify the IP address of the storage complex.
Click Ok. The storage complex that you added is available for selection on the Storage
Complexes main page. Alternatively, you can define a storage complex, and then add the
storage unit from the Storage Unit panel.
The configuration commands will only be applied after exporting the resultant configuration to
an unconfigured DS8000 storage unit.
In the Select Action pull-down we can select Default or Create New. We select Create
New. The next window contains the CMUG00098W warning message; see Figure 13-6.
The CMUG00098W warning message is indicating that two configuration files are open:
– The Default file that we did not select (see the entry in State column of Figure 13-5)
– The new configuration file that we selected to be created
Do not choose Continue. Choose OK to close the Default file and to create the new
configuration file. The successful creation of the new configuration file is shown in
Figure 13-7.
The next panel will be the Long running task summary: Simulated panel, that shows the
task Finished. We select Manage configuration files and we are returned to the starting
Manage configuration files: Simulated panel; see Figure 13-8, In this panel we see that the
Default configuration file is Closed and the new file is Open.
Now, with the Default configuration file Closed and the new file Open we proceed.
3. Select Manage Hardware.
4. Select Storage Unit.
The steps above opened a new configuration file while the steps to come will import a new
configuration from an existing DS8300 —and the configuration information will be placed into
this newly opened file.
When the Storage units: Simulated panel is presented, select Select Action and the screen
will appear as shown in Figure 13-9.
6. In the Identify storage complex panel we enter the IP address and click Next.
7. The next panel is the Select storage unit to import; see Figure 13-11. Here we select the
storage unit of interest and request all the data possible to be imported.
Click Next.
8. The result will be the general information panel, as shown in Figure 13-12 on page 232. In
this figure we have already made an entry in the Description field, which will appear as
blank and you can make a brief entry. The Nickname field can also be edited.
9. A Verification panel is then displayed; see Figure 13-13. Here you will need to verify the
information. Note that the configuration that is going to be imported is from a DS8300
(model 922) with advanced functions installed, the total capacity is just greater than 21 TB,
and there are 16 I/O adapters.
Verify the information and click Finish.
The next panel will be a long running task panel similar to that shown in Figure 13-14 on
page 233 —this figure is an example of when another configuration was being imported;
thus, the resource and the date are different than the specific import that we are using in
our ongoing example. This example shows that the import activity is 56 percent complete.
We simply wait until the state changes to finished.
After some period of time the import activity will complete and a panel similar to that shown in
Figure 13-15 will be presented, indicating that the import has completed. Note that the
example import activity of the 28 TB machine took 2 minutes to complete.
We have now successfully completed the import of a configuration from an existing DS8300.
Next select Manage Hardware and then Storage image. The panel presented will appear
similar to that shown in Figure 13-18 on page 235. Select the storage image to examine by
checking the appropriate box. Note that the only available storage image is serial number
00011. This implies the storage unit serial number is 00010. These are artificial numbers
supplied by the Simulated manager. This is the image that we want to examine.
Now select Configure storage and then Arrays. The shown in Figure 13-19 will display.
Here you select the storage image. When the storage image is selected the panel will
automatically change to that shown in Figure 13-20. The panel displays 16 arrays.
Ranks and extent pools can be viewed if you follow a very similar procedure to that just
outlined for arrays. It is also possible to see the configured ports, as shown in Figure 13-21.
Select Manage hardware, then Storage images, Configure I/O ports in the Select Action
pull-down. For this configuration 64 ports are displayed with the appropriate details. Also
shown in Figure 13-21, on the left of the figure are selections for:
Open systems Volumes
Open systems Volume groups
System z LCUs
System z Volumes
The above discussion and figures provide significant details for importing an existing
configuration and examining the details of that imported configuration.
Simulated panel is presented, select Select Action and the screen shown in Figure 13-22
will display.
The Define properties panel will be displayed. Enter a nickname for the storage complex and
an optional description, as shown in Figure 13-24 on page 238.
Specify the machine model type for the storage unit. If you specify an LPAR machine you will
have the option to define each LPAR separately. Enter a nickname for the storage unit and an
optional description. You can also specify which storage complex this storage unit is
associated with:
1. Click Next.
2. Enter the number of packs of each DDM type. Click Add. Repeat for each DDM type.
3. Specify the licensed functions that you have purchased. Remember that you need to
process the authorized functions, including the operating environment license, before you
can configure the storage.
4. Click Next.
5. Specify the number of I/O adapters (cards) of each type. Each I/O adapter is four ports for
the FCP/FICON cards, and two ports for each ESCON card. Click Add.
6. Click Next.
7. Review the Verification panel. You can click Back to make any changes, or Finish to
accept the configuration.
that you specify does not match the installed configuration, you will not be able to apply the
logical configuration to the HMC.
Once you have imported the file you can review the configuration:
1. From the DS Storage Manager select Simulated Manager.
2. Select Storage Units.
3. Select Properties from the drop-down list.
You can use the configuration files function to create cloned logical configurations for identical
storage units. You could also use the configuration files if you have an environment where you
need to completely replace a configuration.
For each configuration task (for example, creating an array) the process will guide you
through different panels where you have to enter the necessary information. During this you
have the possibility to go back to do modifications or cancel the process. At the end of each
process you will get a verification panel where you can verify the information you entered
before you submit the task.
3. Select Configure IO Ports from the drop-down menu, as shown in Figure 13-25
4. The Configure IO Ports panel is displayed; see Figure 13-26.
Here you select the ports that you want to format as FcSf, FC-AL, or FICON, and then
select the port format from the drop-down menu, as shown in Figure 13-26. The ports are
identified by their location: Rack - I/O enclosure - Card - Port.
You will get a message warning you that if hosts are currently accessing these ports and
you reformat them, it is possible that the hosts will lose access to the ports. This is
because this step in the configuration is when you are selecting whether a port is to be
FICON or FCP.
5. You can repeat this step to format all ports to their required function.
To find out which hosts are logged into the system you use the Query Host WWPNs panel;
see Figure 13-27.
This panel can be used to debug host access and switch configuration issues.
Figure 13-28 shows you existing host systems definitions in the Host systems panel of the
Real-time manager.
From this panel you can select one or more hosts and then select from the Select Action
pull-down menu a specific task (for example, Modify). To create a new host select Create
from the Select Action pull-down menu. The following process will guide you through the host
configuration; see Figure 13-29.
In the General host information panel, you have to enter the following information:
Type: The host type; in our example we create a pSeries host. The pull-down menu gives
you a list of types you can select.
Nickname: Name of the host.
Description: Optionally, you can give a description of the host, for example, the TCP/IP
address of the host or the location of the host.
When you have entered the needed information click Next to define the host ports
The Define host attachments panel will be displayed; see Figure 13-30. In this panel you to
enter the following information:
Quantity of WWPN: The number of Fibre Channel adapters from your host from that you
want to access the DS8000.
Quantity of volume groups: The number of VGs you want to assign to the host.
Attachment Port Type: You have to specify if the host is attached over a FC Switch fabric
(P-P) or direct FC arbitrated loop to the DS8000.
Group ports to share a common set of volumes: If you check this box it means that the
adapters can access the same volumes from the DS8000.
If you are using the Simulated manager, you can optionally specify the WWPN.
You can repeat this for multiple groups of adapters, for example, if you want to create two
adapter definitions for a host. The Defined host ports list will be updated with the information.
In this example we defined two FC adapters in a FC switch fabric configuration and the
adapters are grouped.
The Define host WWPN panel will be displayed; see Figure 13-31.
In the Define host WWPN panel, you have for each FC adapter a field where you have to
enter the World Wide Port Name (WWPN) of the adapter. On the previous panel we defined
that our host has two FC adapters; therefore, we have on the Define host WWPN panel two
fields to enter the WWPNs from the FC adapters.
If you are using the Simulated manager you can optionally specify the WWPN for each
attachment. Alternatively, you can leave this blank (or input a dummy value) and modify the
definition to retrieve the WWPN from a pull-down that will be populated once the host is
attached.
Click Next. The Specify Connection panel will be displayed; see Figure 13-32.
In the Select storage images box, all storage images from the storage unit are listed. Select
the storage images you want to access from the server and click Apply Assignment.
When complete, click Next. This takes you to the Verification panel; see Figure 13-33.
In the Verification panel you have to check the information you entered during the process. If
you want to make modifications select Back or you can Cancel the process. After you verify
the information, click Finish to create the host system.
If you need to make changes to a host system definition you can go to the Host system panel,
select the host system, and then select Modify from the drop-down menu.
You can see in the Arrays panel in Figure 13-34 that already more than ten arrays are
created. To create a new array, select Create from the drop-down menu.
You will be guided through the process to build a new array; see Figure 13-35
Create arrays automatically: The system will choose an array site to build the array. You
specify the quantity of arrays and RAID type. You can optionally specify the format of the
rank and add the array to a rank.
Create custom arrays: Then you have the following panel to choose the RAID type and the
array site to create the array (see Example 13-36).
In this panel you have to select the RAID type (it can be RAID 5 or RAID 10), and you have to
choose the array site from which the array will be built. In this example we create a RAID 5
array on array site S3. If you create multiple arrays at one time, the arrays will be created over
array sites distributed across the DAs. The array number is created by default.
In the Add arrays to ranks panel you have the possibility of adding the array to a rank.
If you want the array you are creating to be added to a rank then you have to select Add
these arrays to ranks, and also specify whether it should be a FB (fixed block) rank for open
system hosts or a CKD (count key data) rank for System z hosts; see Figure 13-37.
If you do not select Add these arrays to ranks you have to later assign the array to a rank.
Click Next. This will take you to the Verification panel; see Figure 13-38.
In this panel you can verify your specifications for the array. If everything is correct click
Finish to create the array. You can check the progress of the array creation by using the Long
Running Task monitor.
To delete arrays use the Delete option from the drop-down menu on the main Arrays panel. If
there are volumes allocated on the arrays to be deleted a warning is issued. However, you
can still proceed with deletion.
You can see that already more than ten are created. To create another rank you can choose
from the Select Action pull-down menu create and click Go. This will start the process to
create a rank.
First you have to select the array, or arrays, from which you want to build the ranks. In our
example only one array is available that we can choose; see Figure 13-40.
The Define rank properties panel will display; see Figure 13-41 on page 249.
In the Define rank properties panel you decide if you want to create a FB (fixed block) rank for
open system servers or a CKD (Count Key Data) rank for System z server. In our example we
select Storage type FB.
Optionally, you can select the extent pool to which you want to assign the rank and click Next
to get the Verification panel; see Figure 13-42. If you have not yet created extent pools you
can leave this step until later.
In the Verification panel (Figure 13-43) you can verify your specifications for the rank. If
everything is correct click Finish to create the rank. The rank number is generated by default.
You can monitor the progress of the Create rank task using the Long Running Task Monitor.
You can delete a rank using the Delete option from the drop-down menu in the Rank main
panel. If there are any volumes allocated on the rank this will result in a warning message, but
volumes will be deleted if the rank is deleted. You can also use the Modify option to change
the format of a rank, or remove a rank from an extent pool. You can only modify the properties
of a rank if there are no extents allocated.
To create a new extent pool select Create from the Select Action pull-down list. The Definition
method panel is displayed; see Figure 13-45.
In the Definition method panel you select the definition method. You can choose between:
Create extent pool automatically based on storage requirements: You specify the amount
of storage, the RAID type, and the format (FB or CKD) you need and the system will
automatically use unassigned ranks of the type you specify to fulfill your requirements.
Create custom extent pool: You specify the ranks that you assign to the extent pool.
The Define properties panel is displayed; see Figure 13-46. In this panel we enter:
Nickname: Name of the extent pool.
Storage type: Select FB for open systems hosts or CKD for System z servers.
RAID type: You can select RAID 5 or RAID 10.
Server: ranks acquire server affinity from being added into an extent pool. It is
recommended that you balance ranks between server 0 and server 1.
The Select ranks panel is displayed; see Figure 13-47. here you select the ranks that you
want to add to the extent pool.
In our example there are only two RAID 5, FB ranks defined so you can only select from these
ranks. We recommend that you create extent pools with one rank for maximum management
of your performance environment; however, you are required to create a minimum of one
extent pool for each disk format (FB or CKD). If you want to manage your environment with
the least number of extent pools you should have one extent pool for each server for FB and
CKD data.
In the next panel (Figure 13-48) you can specify Reserved storage, which is the amount of
storage that you choose to reserve for later use. Capacity that is reserved cannot be used to
create volumes; volumes may only be allocated in reserved space once the space is explicitly
released using the Modify option. You might use reserved space to keep capacity in the
extent pool free for planned application growth. Usually you will enter 0, which means that you
can use all of the storage.
In this panel you can verify your specifications for the extent pool. If everything is correct click
Finish to create the extent pool.
If you create the extent pools before you create the ranks, you can add the ranks to the
created extent pools as part of rank creation. Alternatively, you can go to the Rank main panel
and select Add rank to extent pool from the drop-down menu.
Choose Create from the Select Action pull-down menu. The Select extent pool panel will
display; see Figure 13-51.
In the Select extent pool panel choose the previously created extent pool from which you want
to create the volume (you can choose only one extent pool). If you have not yet created extent
pools you can do so by clicking the button to enter the Create new extent pool wizard.
When finished with the previous task, you then have to define the volume characteristics.
In the Define volume characteristics panel (see Figure 13-52) the fields are:
Volume type: You can choose Standard Open System volumes in either binary (DS sizes)
or decimal (ESS sizes) for compatibility with ESS. In general, the DS sizes should be fine.
You can select either protected or unprotected iSeries volumes. Select unprotected iSeries
volumes if you are going to use iSeries mirroring.
Select volume groups: You can select one or more volume groups to which you want to
assign the volumes. If you choose no volume group you can assign the volumes later to a
volume group by using the Modify option. Alternatively, you can click Create new group to
enter the Create volume group wizard.
When complete click Next to define the volumes properties; see Figure 13-53.
In the Define volume properties panel you have to enter the following information:
Quantity: The number of volumes you want to create. The calculator will tell you how many
volumes of your chosen size will fit in the available space in the extent pool.
Size: Size of the volumes in GB (binary or decimal). The minimum allocation is 0.1 GB;
however, this will consume an entire 1 GB extent. The maximum LUN size is 2 TB. iSeries
volumes can only be created in the sizes supported by the operating system; these are
selectable from a drop-down menu.
Select LSSs for volumes: If you select this check box you can specify the LSS for the
volumes. Only the LSSs that are available will be displayed. In this example we have only
even LSSs displayed because of selecting an extent pool associated with server 0.
You can assign the volumes to a specific LSS. This can be important if you want to use Copy
Services. You can have a maximum of 256 volumes in each LSS.
In this example we create two volumes with 5 GB each assigned to LSS 10; see Figure 13-53.
In this example the two volumes will get the names MetroM0001 and MetroM0002; see
Figure 13-54. You can see this also in the verification panel; see Figure 13-55.
In the Verification panel you check the information you entered during the process. If you want
to make modifications select Back or you can Cancel the process. After you verify the
information click Finish to create the volume.
4. To create a new volume group select Create from the Action pull-down menu; see
Figure 13-56.
5. In the Define volume group properties panel (see Figure 13-57) enter the nickname for the
volume group and select the host type from which you want to access the volume group. If
you select one host (for example, System p), all other host types with the same addressing
method will be automatically selected. This does not affect the functionality of the volume
group; it will support the host type selected.
6. Select the host attachment for your volume group. You can select only one host
attachment. See Figure 13-58.
In the Select volumes for group panel you have to select the volumes that should be included
in the volume group; see Figure 13-59.
If you have to select a large number of volumes, you can specify a filter so that only these
volumes are displayed in the list, and then you can select all.
In the Verification panel you have to check the information you entered during the process. If
you want to make modifications select Back, or you can Cancel the process. After you verify
the information click Finish to create the host system attachment.
Already four LCUs are created. To create one or more new LCUs choose Create from the
Select the Action pull-down menu; see Figure 13-61.
In Select from available LCUs panel you can select the LCU IDs you want to create; see
Figure 13-62.
In this example we create LCUs 05 and 06. When finished click Next.
In the Define LCU properties panel (see Figure 13-63) you can define the LCU properties.
SSID: Enter the Subsystem ID (SSID) for the LCU. If you create multiple LCUs in one step
the SSIDs will be incremented.
LCU type: Select the LCU type you want to create (3990 Mod.3, 3990 Mod. 3 for TPF, or
3990 Mod. 6).
Concurrent copy session time-out (sec.): The time in seconds that any logical device on
this LCU in a concurrent copy session stays in a long busy state before suspending a
concurrent copy session.
z/OS Global Mirror Session time-out: The time in seconds that any logical device in an
XRC session stays long busy before suspending an XRC session. The long busy occurs
because the data mover has not off-loaded data when the logical device (or XRC session)
is no longer able to accept additional data. You can change the default time (in seconds)
by highlighting it and entering a new time.
On the Verification panel you have to check the information you entered during the process. If
you want to make modifications select Back, or you can Cancel the process. After you verify
the information click Finish to create the LCU.
The Volumes - zSeries panel displays; see Figure 13-65 on page 262.
To create new System z volumes select Create from the Select Action pull-down menu.
There is another option in the drop-down menu to Define Address Allocation Policy, which
allows you to set defaults for the address group for the start address and whether addresses
will be incremented or decremented. You can override these defaults when you create
volumes, or alternatively you can specify these parameters when you create the volumes.
In the Select extent pool panel (see Figure 13-66) select the extent pool in which you want to
create the volumes. You can select only one extent pool in each step, although you can create
volumes across multiple LSSs in one step. To create volumes in multiple extent pools you
must repeat this step.
You then have to define the base volume characteristics; see Figure 13-67.
Click Next to define the base volume properties; see Figure 13-68.
Quantity: The number of base volumes you want to create. The calculator will show how
many volumes of your selected size you can create in the extent pool that you selected.
The calculator also tells you how many addresses are available in the LCUs that you have
selected.
Base start address: The address of the first volumes you want to create. This will default at
the value you specify in the address policy definition.
Ascending/Descending: Select the addressing order for the base volumes. This will default
at the value you specify in the address policy definition.
In the Create volume nicknames we create the nicknames for the volumes; see Figure 13-69.
In this example we create the base volumes VOL0001, VOL0002... VOL0064. You can specify
prefix and suffixes up to a total of 16 characters.
You have the option to use a hexadecimal addressing sequence typical of System z
addressing if you check the box Use hexadecimal sequence.
2. In the Create zSeries volume panel (see Figure 13-70) select the volumes for which you
want to create alias volumes, for which you can use the Select all check box, and make the
selections for the following:
– Starting address: In this field you have to enter the first alias address.
– Ascending/Descending: Select the addressing order for the aliases.
– Aliases/Per Volumes: In these two fields you have to enter how many aliases you want
to create for each selected base volume. This creates a ratio between base and alias
addresses that will be applied to all the volumes selected. The ratio can be multiple
aliases for each base address, or multiple base addresses to each alias; however, only
whole numbers and evenly divisible ratios are acceptable, for example:
• One alias: Two base addresses
• Two aliases: One base address
• Three aliases: Six base addresses
• But not three aliases: Two base addresses
In this example we are creating two aliases for each base address.
Click Add aliases and click Next to go to the verification panel; see Figure 13-71 on
page 266.
Verify that you entered the correct information before you click Finish.
14
In this chapter we discuss the use of the DS CLI for storage configuration of the DS8000, not
for Copy Services configuration. For Copy Services configuration in the DS8000 using the DS
CLI refer to the following books:
IBM System Storage DS8000: Command-Line Interface User´s Guide, SC26-7916
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787
Note: The DS CLI is used for both the DS8000 and the DS6000. In most cases, the
commands operate in exactly the same way for both storage units. If you see a command
being run on a DS6000, it will probably work in exactly the same manner on a DS8000.
There number of commands that are unique to either disk subsystem are very few.
Example 14-1 Displaying a list of all commands in DS CLI using the help command
dscli> help
applykey lsframe mkpprc setdialhome
chckdvol lshba mkpprcpath setflashrevertible
chextpool lshostconnect mkrank setioport
chfbvol lshosttype mkremoteflash setoutput
chhostconnect lshostvol mksession setplex
chlcu lsioport mkuser setremoteflashrevertible
chlss lskey mkvolgrp setsim
chpass lslcu offloadss setsmtp
chrank lslss pausegmir setsnmp
chsession lsportprof pausepprc setvpn
chsi lspprc quit showarray
chsp lspprcpath restorevolaccess showarraysite
chsu lsproblem resumegmir showckdvol
chuser lsrank resumepprc showcontactinfo
chvolgrp lsremoteflash resyncflash showextpool
clearvol lsserver resyncremoteflash showfbvol
closeproblem lssession reverseflash showgmir
commitflash lssi revertflash showgmircg
commitremoteflash lsstgencl revertremoteflash showgmiroos
dscli lssu rmarray showhostconnect
exit lsuser rmckdvol showioport
failbackpprc lsvolgrp rmextpool showlcu
failoverpprc managehostconnect rmfbvol showlss
freezepprc managepwfile rmflash showpass
help mkaliasvol rmgmir showplex
lsaddressgrp mkarray rmhostconnect showrank
lsarray mkckdvol rmlcu showsi
lsarraysite mkesconpprcpath rmpprc showsp
lsavailpprcport mkextpool rmpprcpath showsu
lsckdvol mkfbvol rmrank showuser
lsda mkflash rmremoteflash showvolgrp
lsddm mkgmir rmsession testcallhome
lsextpool mkhostconnect rmuser unfreezeflash
lsfbvol mklcu rmvolgrp unfreezepprc
lsflash mkpe setcontactinfo ver
dscli>
When configuring a DS8000 with the DS CLI, you are required to include the machine’s ID in
nearly every command that is issued. If we do not want to type this ID in after each command,
then you need to change the DS CLI profile. If you enter the serial number of the machine and
the HMC’s network address into this profile, you will not have to include this field in each
command.
4. Now you have notepad opened with the DS CLI profile. There are four lines you could
consider adding. Examples of these are shown in bold in Example 14-3.
devid: IBM.2107-75ABCDE
hmc1: 10.0.0.250
username: admin
password: passw0rd
5. Save the changed file as a new file and close notepad. You can then reference the new
profile using the -cfg parameter when starting DS CLI.
Attention: The default profile file created when you install DS CLI will potentially be
replaced every time you install a new version of DS CLI. It is a better practice to open the
default profile and then save it as a new file. You can then create multiple profiles and
reference the relevant profile file using the -cfg parameter.
Adding the serial number using the devid parameter, and the HMC IP address using the hmc1
parameter is highly recommended. Adding the username and password parameters will
certainly simplify your DS CLI startup, but it is less recommended. This is because a
password saved in a profile file is saved in clear text. Anyone who has access to that file can
then read the password.
Important: Take care if adding multiple devid and HMC entries. Only one should be
uncommented (or more literally, unhashed) at any one time. If you have multiple hmc1 or
devid entries, the DS CLI uses the one closest to the bottom of the profile.
In Example 14-5 we set two I/O ports to the FICON topology and then check the results.
The first task, set the I/O ports, has already been discussed in 14.2, “Configuring the I/O
ports” on page 270. The second task, install the license keys, has already been discussed in
11.2.4, “Applying activation codes using the DS CLI” on page 209.
Attention: Remember that an array for a DS8000 can only contain one array site and a
DS8000 array site contains eight disk drive modules (DDMs).
In Example 14-6, we can see that there are four array sites and that we can therefore create
four arrays.
We can now issue the mkarray command to create arrays, as in Example 14-7. You will notice
that in this case we have used one array site (in the first array, S1) to create a single RAID 5
array. If we wished to create a RAID 10 array, we would have to change the -raidtype
parameter to 10 (instead of 5).
We can now see what arrays have been created by using the lsarray command; see
Example 14-8.
Example 14-8 shows the result of the lsarray command. We can see the type of RAID array
and the number of disks that are allocated to the array (in this example 6+P+S, which means
the usable space of the array is 6 times the DDM size), as well as the capacity of the DDMs
that are used and which array sites were used to create the arrays.
Once we have created all the ranks, we do an lsrank command. This command will display
all the ranks that have been created, which server the rank is attached to, the RAID type, and
the format of the rank—whether it is Fixed Block (FB) or Count Key Data (CKD).
Example 14-9 shows the mkrank commands and the result of a succeeding lsrank -l
command.
Example 14-9 Creating and listing ranks with mkrank and lsrank
dscli> mkrank -array A0 -stgtype fb
Date/Time: 27 October 2005 21:31:16 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
CMUC00007I mkrank: Rank R0 successfully created.
dscli> mkrank -array A1 -stgtype fb
Date/Time: 27 October 2005 21:31:16 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
CMUC00007I mkrank: Rank R1 successfully created.
dscli> lsrank -l
Date/Time: 27 October 2005 21:32:31 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
ID Group State datastate Array RAIDtype extpoolID extpoolnam stgtype exts usedexts
=======================================================================================
R0 - Unassigned Normal A0 5 - - fb 773 -
R1 - Unassigned Normal A1 5 - - fb 773 -
For ease of management, we create empty extent pools relating to the type of storage that is
in this pool. For example, create an extent pool for high capacity disk, create another for high
performance, and, if needed, extent pools for the CKD environment. For high capacity, you
would consider using 300 GB 10k rpm DDMs, while for high performance you might consider
73 GB 15k rpm DDMs.
It is also a good idea to note to which server the extent pool has an affinity.
Example 14-10 shows an example of how we could divide your machine. Now in
Example 14-6 on page 271 we only had four array sites, so clearly we would need more
DDMs to support this many extent pools.
Note that the mkextpool command forces you to name the extent pools. In Example 14-11 we
first create empty extent pools using mkextpool. We then list the extent pools to get their IDs.
Then we attach a rank to an empty extent pool using the chrank command. Finally we list the
extent pools again using lsextpool and note the change in capacity of the extent pool.
Example 14-11 Extent pool creation using mkextpool, lsextpool, and chrank
dscli> mkextpool -rankgrp 0 -stgtype fb FB_high_0
Date/Time: 27 October 2005 21:42:04 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
CMUC00000I mkextpool: Extent Pool P0 successfully created.
dscli> mkextpool -rankgrp 1 -stgtype fb FB_high_1
Date/Time: 27 October 2005 21:42:12 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
CMUC00000I mkextpool: Extent Pool P1 successfully created.
dscli> lsextpool
Date/Time: 27 October 2005 21:49:33 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
Name ID stgtype rankgrp status availstor (2^30B) %allocated available reserved numvols
===========================================================================================
FB_high_0 P0 fb 0 below 0 0 0 0 0
FB_high_1 P1 fb 1 below 0 0 0 0 0
dscli> chrank -extpool P0 R0
Date/Time: 27 October 2005 21:43:23 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
CMUC00008I chrank: Rank R0 successfully modified.
dscli> chrank -extpool P1 R1
Date/Time: 27 October 2005 21:43:23 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
CMUC00008I chrank: Rank R1 successfully modified.
dscli> lsextpool
Date/Time: 27 October 2005 21:50:10 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
Name ID stgtype rankgrp status availstor (2^30B) %allocated available reserved numvols
===========================================================================================
FB_high_0 P0 fb 0 below 773 0 773 0 0
FB_high_1 P1 fb 1 below 773 0 773 0 0
After having assigned a rank to an extent pool, we should be able to see this when we display
the ranks. In Example 14-12 on page 274 we can see that rank R0 is assigned to extpool P0.
Example 14-12 Displaying the ranks after assigning a rank to an extent pool
dscli> lsrank -l
Date/Time: 27 October 2005 22:08:42 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
ID Group State datastate Array RAIDtype extpoolID extpoolnam stgtype exts usedexts
===================================================================================
R0 0 Normal Normal A0 5 P0 FB_high_0 fb 773 0
R1 1 Normal Normal A1 5 P1 FB_high_1 fb 773 0
In Example 14-13, we have created eight volumes, each with a capacity of 10 GB. The first
four volumes are assigned to rank group 0 and the second four are assigned to rank group 1.
Looking closely at the mkfbvol command used in Example 14-13, we see that volumes
1000–1003 are in extpool P0. That extent pool is attached to rank group 0, which means
server 0. Now rank group 0 can only contain even numbered LSSs, so that means volumes in
that extent pool must belong to an even numbered LSS. The first two digits of the volume
serial number are the LSS number, so in this case, volumes 1000–1003 are in LSS 10.
For volumes 1100–1003 in Example 14-13, the first two digits of the volume serial number are
11, which is an odd number, which signifies they belong to rank group 1. Also note that the
-cap parameter determines size, but because the -type parameter was not used, the default
size is a binary size. So these volumes are 10 GB binary, which equates to 10,737,418,240
bytes. If we used the parameter -type ess, then the volumes would be decimally sized and
would be a minimum of 10,000,000,000 bytes in size.
Finally, in Example 14-13 we named the volumes using the naming base: high_fb_0_#h. The
#h means to use the hexadecimal volume number as part of the volume name. This can be
seen in Example 14-14 on page 275, where we list the volumes that we have created using
the lsfbvol command. We then list the extent pools to see how much space we have left after
the volume creation.
Example 14-14 Checking the machine after creating volumes, by using lsextpool and lsfbvol
dscli> lsfbvol
Date/Time: 27 October 2005 22:28:01 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
Name ID accstate datastate configstate deviceMTM datatype extpool cap (2^30B) cap (10^9B)
===========================================================================================================
high_fb_0_1000 1000 Online Normal Normal 2107-922 FB 512 P0 10.0 -
high_fb_0_1001 1001 Online Normal Normal 2107-922 FB 512 P0 10.0 -
high_fb_0_1002 1002 Online Normal Normal 2107-922 FB 512 P0 10.0 -
high_fb_0_1003 1003 Online Normal Normal 2107-922 FB 512 P0 10.0 -
high_fb_1_1100 1100 Online Normal Normal 2107-922 FB 512 P1 10.0 -
high_fb_1_1101 1101 Online Normal Normal 2107-922 FB 512 P1 10.0 -
high_fb_1_1102 1102 Online Normal Normal 2107-922 FB 512 P1 10.0 -
high_fb_1_1103 1103 Online Normal Normal 2107-922 FB 512 P1 10.0 -
dscli> lsextpool
Date/Time: 27 October 2005 22:27:50 IBM DSCLI Version: 5.1.0.204 DS: IBM.2107-7503461
Name ID stgtype rankgrp status availstor (2^30B) %allocated available reserved numvols
===========================================================================================
FB_high_0 P0 fb 0 below 733 5 733 0 4
FB_high_1 P1 fb 1 below 733 5 733 0 4
Important: For the DS8000, the LSSs can be ID 00 to ID FE. The LSSs are in address
groups. Address group 0 is LSS 00 to 0F, address group 1 is LSS 10 to 1F, and so on. The
moment you create an FB volume in an address group, then that entire address group can
only be used for FB volumes. Be aware of this when planning your volume layout in a
mixed FB/CKD DS8000.
=====================================================
AMDLinuxRHEL AMD - Linux RHEL LUNPolling 512
AMDLinuxSuse AMD - Linux Suse LUNPolling 512
AppleOSX Apple - OSX LUNPolling 512
Fujitsu Fujitsu - Solaris LUNPolling 512
HpTru64 HP - Tru64 LUNPolling 512
HpVms HP - Open VMS LUNPolling 512
LinuxDT Intel - Linux Desktop LUNPolling 512
LinuxRF Intel - Linux Red Flag LUNPolling 512
LinuxRHEL Intel - Linux RHEL LUNPolling 512
LinuxSuse Intel - Linux Suse LUNPolling 512
Novell Novell LUNPolling 512
SGI SGI - IRIX LUNPolling 512
SanFsLinux - Linux/SanFS LUNPolling 512
Sun SUN - Solaris LUNPolling 512
VMWare VMWare LUNPolling 512
Win2000 Intel - Windows 2000 LUNPolling 512
Win2003 Intel - Windows 2003 LUNPolling 512
iLinux IBM iSeries - iLinux LUNPolling 512
pLinux IBM pSeries - pLinux LUNPolling 512
Having determined the host type, we can now make a volume group. In Example 14-16 the
example host type we chose is AIX, and checking Example 14-15, we can see the address
discovery method for AIX is scsimask.
In this example we added volumes 1000 to 1002 and 1100 to 1102 —we did this to spread the
workload evenly across the two rank groups. We then listed all available volume groups using
lsvolgrp. Finally, we listed the contents of volume group V11, since this was the volume
group we created.
Clearly we may also want to add or remove volumes to this volume group at a later time. To
achieve this we use chvolgrp with the -action parameter. In Example 14-17 on page 277 we
added volume 1003 to the volume group V11. We display the results, and then removed the
volume.
Attention: Not all operating systems can deal with the removal of a volume. Consult your
operating system documentation to determine the safest way to remove a volume from a
host.
In Example 14-18 we create a single host connection —that represents one HBA in our
example AIX host. We use the -hosttype parameter using the hosttype we determined in
Example 14-15 on page 275. We allocated it to volume group V11. At this point —provided
the SAN zoning is correct— the host should be able to see the LUNs in volume group V11.
Note that you can also use just -profile instead of -hosttype. However, this is not
recommended. If you use the -hosttype parameter it will actually invoke both parameters
(-profile and -hosttype), where using just -profile will leave the -hosttype column
un-populated.
There is also the option in the mkhostconnect command to restrict access to only certain I/O
ports. This is done with the -ioport parameter. Restricting access in this way is usually not
necessary. If you wish to restrict access for certain hosts to certain I/O ports on the DS8000,
do this through zoning on your SAN switch.
When creating hosts you can specify the -portgrp parameter. By using a unique port group
number for each attached server, you can easily detect servers with multiple HBAs.
In Example 14-19 we have six host connections. By using the port group number we can see
there are three separate hosts, each with two HBAs. Port group 0 is used for all hosts that do
not have a port group number set.
In this section we give examples for several operating systems. In each example we assign
some storage unit volumes (either DS8000 or DS6000) to an open systems host. We install
DS CLI on this host. We log on to this host and start DS CLI. It does not matter which HMC or
SMC we connect to with the DS CLI. We then issue the lshostvol command.
Important: The lshostvol command communicates only with the operating system of the
host on which the DS CLI is installed. You cannot run this command on one host to see the
attached disks of another host.
Example 14-22 lshostvol on an HP-UX host that does not use SDD
dscli> lshostvol
Date/Time: November 16, 2005 4:03:25 AM GMT IBM DSCLI Version: 5.0.4.140
Disk Name Volume Id Vpath Name
==========================================
c38t0d5 IBM.2107-7503461/1105 ---
c38t0d6 IBM.2107-7503461/1106 ---
Example 14-24 lshostvol on a Solaris host that does not have SDD
dscli> lshostvol
Date/Time: November 10, 2005 3:58:29 PM MET IBM DSCLI Version: 5.1.0.204
Disk Name Volume Id Vpath Name
==========================================
c6t1d0 IBM-2107.7520781/4200 ---
c6t1d1 IBM-2107.7520781/4201 ---
c7t2d0 IBM-2107.7520781/4200 ---
c7t2d1 IBM-2107.7520781/4201 ---
Example 14-25 lshostvol on a Windows host that does not use SDD
dscli> lshostvol
Date/Time: 11. November 2005 12:02:26 CET IBM DSCLI Version: 5.1.0.204
Disk Name Volume Id Vpath Name
==========================================
Disk0 IBM.1750-1300247/1400 ---
Disk1 IBM.1750-1300247/1401 ---
Disk2 IBM.1750-1300247/1402 ---
Disk3 IBM.1750-1300247/1403 ---
You do not have to create volume groups or host connects for CKD volumes. Provided there
are I/O ports in FICON mode, access to CKD volumes by FICON hosts will be granted
automatically.
So first we must use the mklcu command. The format of the command is:
mklcu -qty XX -id XX -ssXX
Then, to display the LCUs that we have created, we can use the lslcu command.
In Example 14-28 on page 282, we create two LCUs using mklcu, and then list the created
LCUs using lslcu. Note that by default, the LCUs that were created are 3990-6.
Also note that because we created two LCUs (using the parameter -qty 2), the first LCU
being ID 00 (an even number) is in address group 0, which equates to rank group 0; while the
second LCU being ID 01 (an odd number) is in address group 1, which equates to rank group
1. By placing the LCUs into both address groups we maximize performance by spreading
workload across both rank groups of the DS8000.
Note: For the DS8000, the CKD LCUs can be ID 00 to ID FE. The LCUs will fit into one of
16 address groups. Address group 0 is LCUs 00 to 0F, address group 1 is LCUs 10 to 1F,
and so on. If you create a CKD LCU in an address group, then that address group cannot
be used for FB volumes. Likewise, if there were, for instance, FB volumes in LSS 40 to 4F
(address group 4), then that address group cannot be used for CKD. Be aware of this
when planning the volume layout in a mixed FB/CKD DS8000.
The major difference to note here is that the capacity must be equivalent to that of a 3390
model 3 or a multiple of this capacity (3339 cylinders) so that there is very little or no space
wasted.
Remember we can only create CKD volumes using LCUs that we have already created. From
our examples, trying, for instance, to make volume 0200 will fail with the same message seen
in Example 14-27 on page 281. This is because we only created LCU IDs 00 and 01,
meaning all CKD volumes must be in the address range 00xx (LCU ID 00) and 01xx (LCU ID
01).
You also need to be aware that volumes in even numbered LCUs must be created from an
extent pool that belongs to rank group 0, while volumes in odd numbered LCUs must be
created from an extent pool in rank group 1.
In Example 14-30 we have created a file called samplebat.bat. Because it is a Windows batch
file and we are issuing individual DS CLI commands, it can also contain any command that
can be executed in a Windows command prompt.
Having creating the BAT file, we can run it and display the output file. An example is shown in
Example 14-31. We run the batch file samplebat.bat and the command output is displayed.
D:\>
In Example 14-32 we have the contents of a DS CLI script file. Note that it only contains DS
CLI commands, though comments can be placed in the file using a hash (#). One advantage
of using this method is that scripts written in this format can be used by the DS CLI on any
operating system into which you can install DS CLI.
In Example 14-33, we start the DS CLI using the -script parameter and specifying the name
of the script that contains the commands from Example 14-32.
C:\Program Files\ibm\dscli>
Part 4
Part 4 Host
considerations
In this part we discuss the specific host considerations that you may need for implementing
the DS8000 with your chosen platform. The following host platforms are presented:
Open systems considerations
System z considerations
System i considerations
15
Also, some general considerations are discussed at the beginning of this chapter.
For each query, select one storage system and one operating system only; otherwise the
output of the tool will be ambiguous. You will be shown a list of all supported HBAs together
with the required firmware and device driver levels for your combination. Furthermore, you
can select a detailed view for each combination with more information, quick links to the HBA
vendors’ Web pages and their IBM supported drivers, and a guide to the recommended HBA
settings.
already extensive interoperability efforts to develop and deliver products and solutions that
work together with third-party products.
The System Storage Proven Web site provides more detail on the program, as well as the list
of pre-tested configurations:
http://www.ibm.com/servers/storage/proven/index.html
QLogic Corporation
The Qlogic Web site can be found at:
http://www.qlogic.com
QLogic maintains a page that lists all the HBAs, drivers, and firmware versions that are
supported for attachment to IBM storage systems:
http://www.qlogic.com/support/ibm_page.html
Emulex Corporation
The Emulex home page is:
http://www.emulex.com
They also have a page with content specific to IBM storage systems:
http://www.emulex.com/ts/docoem/framibm.htm
JNI/AMCC
AMCC took over the former JNI, but still markets FC HBAs under the JNI brand name. JNI
HBAs are supported for DS8000 attachment to SUN systems. The home page is:
http://www.amcc.com
Atto
Atto supplies HBAs, which IBM supports for Apple Macintosh attachment to the DS8000.
Their home page is:
http://www.attotech.com
They have no IBM storage specific page. Their support page is:
http://www.attotech.com/support.html
IBM in regard to these products. It is beyond the scope of this redbook to list all the vendors
Web sites.
The IBM System Storage DS8000 Host Systems Attachment Guide, SC26-7917, helps you
with the procedures necessary to set up your host in order to boot from the DS8000.
The IBM System Storage Multipath Subsystem Device Driver User’s Guide, SC30-4131, also
helps with identifying the optimal configuration and lists the steps required to boot from
multipathing devices.
Most operating systems, however, cannot deal natively with multiple paths to a single disk.
This puts the data integrity at risk, because multiple write requests can be issued to the same
data and nothing takes care of the correct order of writes.
To utilize the redundancy and increased I/O bandwidth you get with multiple paths, you need
an additional layer in the operating system’s disk subsystem to recombine the multiple disks
seen by the HBAs into one logical disk. This layer manages path failover, should a path
become unusable, and balancing of I/O requests across the available paths.
IBM Multi-path Subsystem Device Driver (SDD) provides load balancing and enhanced data
availability capability in configurations with more than one I/O path between the host server
and the DS8000. SDD performs dynamic load balancing across all available preferred paths
to ensure full utilization of the SAN and HBA resources.
http://www.ibm.com/servers/storage/support/software/sdd/downloading.html
When you click the Subsystem Device Driver downloads link, you will be presented a list of
all operating systems for which SDD is available. Selecting one leads you to the download
packages, the user´s guide, and additional support information.
The user’s guide, the IBM System Storage Multipath Subsystem Device Driver User’s Guide,
SC30-4131, contains all the information that is needed to install, configure, and use SDD for
all supported operating systems.
Note: SDD and RDAC, the multipathing solution for the IBM System Storage DS4000
series, can coexist on most operating systems, as long as they manage separate HBA
pairs. Refer to the DS4000 series documentation for detailed information.
Most of these solutions are also supported for DS8000 attachment, although the scope may
vary. There may be limitations for certain host bus adapters or operating system versions.
Always consult the DS8000 Interoperability Matrix for the latest information.
15.2 Windows
DS8000 supports Fibre Channel attachment to Microsoft Windows 2000/2003 servers. For
details regarding operating system versions and HBA types see the DS8000 Interoperability
Matrix, available at:
http://www.ibm.com/servers/storage/disk/DS8000/interop.html
The support includes cluster service and acts as a boot device. Booting is supported currently
with host adapters QLA23xx (32 bit or 64 bit) and LP9xxx (32 bit only). For a detailed
discussion about SAN booting (advantages, disadvantages, potential difficulties, and
troubleshooting), we highly recommend the Microsoft document Boot from SAN in Windows
Server 2003 and Windows 2000 Server, available at:
http://www.microsoft.com/windowsserversystem/wss2003/techinfo/plandeploy/BootfromSANinWi
ndows.mspx
To ensure optimum availability and recoverability when you attach a storage unit to a
Windows 2000/2003 host system, we recommend setting the Time Out Value value
associated with the host adapters to 60 seconds. The operating system uses the Time Out
Value parameter to bind its recovery actions and responses to the disk subsystem. The value
The value has the data type REG-DWORD and should be set to 0x0000003c hexadecimal
(60 decimal).
In Figure 15-1 you see an example of two disks connected by four paths to the server. You
see two IBM 2107900 SDD Disk Devices as real disks on Windows. The IBM 2107900 SCSI
Disk Device is hidden by SDD. The Disk manager view is shown in Figure 15-2.
Note: New assigned disks will be discovered; if not, go to Disk manager and rescan disks
or to the Device manager and scan for hardware changes.
Example 15-1 is a sample output of the command datapath query device -l. You see that
all paths of the DS8000 will be used —because the DS8000 does not implement the concept
of preferred path, as the DS6000 does.
Total Devices : 2
Another helpful command is datapath query wwpn, shown in Example 15-2. It helps you to
get the World Wide Port Name (WWPN) of your Fibre Channel adapter.
The commands datapath query essmap and datapath query portmap are not available.
information displayed by the datapath query device command, plus the information
displayed in the Windows Disk Management panel and combine them.
In our example of Figure 15-1 on page 292, if we listed the vpaths we could see that SDD
DEV#: 0 has DEVICE NAME: Disk2. We could also see the serial number of the disk is
75065711000, which breaks out as LUN ID 1000 on DS8000 serial 7506571. We then need
to look at the Windows Disk Management panel, an example of which is shown in Figure 15-2
on page 292. From this we can see that Disk 2 is Windows drive letter E. If you can view this
document in color you can see the blue circle in Figure 15-2. SDD DEV#: 1 corresponds to
the red circle around Windows drive letter G.
Now that we have mapped the LUN ID to a Windows drive letter, if driver letter E was no
longer required on this Windows server, we could safely un-assign LUN ID 1000 on the
DS8000 with serial number 7506571, knowing we have removed the correct drive.
Note: The adapter goes to DEGRAD state when there are active paths left on the adapter.
It goes to FAILED state when there are no active paths.
With Windows 2003, MSCS uses target resets. See the Microsoft technical article
Microsoft Windows Clustering: Storage Area Networks at:
http://www.microsoft.com/windowsserver2003/techinfo/overview/san.mspx
Windows Server 2003 will allow for boot disk and the cluster server disks hosted on the
same bus. However, you would need to use Storport miniport HBA drivers for this
functionality to work. This is not a supported configuration in combination with drivers of
other types (for example, SCSI port miniport or Full port drivers).
If you reboot a system with adapters while the primary path is in a failed state, you must
manually disable the BIOS on the first adapter and manually enable the BIOS on the
second adapter. You cannot enable the BIOS for both adapters at the same time. If the
BIOS for both adapters is enabled at the same time and there is a path failure on the
primary adapter, the system will stop with an INACCESSIBLE_BOOT_DEVICE error upon
reboot.
VDS is a set of APIs that uses two sets of providers to manage storage devices. The built-in
VDS software providers enable you to manage disks and volumes at the operating system
level. VDS hardware providers supplied by the hardware vendor enable you to manage
hardware RAID Arrays. Windows Server 2003 components that work with VDS include the
Disk Management Microsoft Management Console (MMC) snap-in; the DiskPart
command-line tool; and the DiskRAID command-line tool, which is available in the Windows
Server 2003 Deployment Kit. Figure 15-3 shows the VDS architecture.
Command-Line Tools:
Disk Management Storage Management
•DiskPart
MMC Snap-in Applications
•DiskRAID
Other
Disk
Subsystem
HDDs LUNs
DS8000/DS6000
Hardware
Microsoft Functionality
Non-Microsoft Functionality
For a detailed description of VDS, refer to the Microsoft Windows Server 2003 Virtual Disk
Service Technical Reference at:
http://www.microsoft.com/Resources/Documentation/windowsserv/2003/all/techref/en-us/W2K3
TR_vds_intro.asp
The DS8000 can act as a VDS hardware provider. The implementation is based on the DS
Common Information Model (CIM) agent, a middleware application that provides a
CIM-compliant interface. The Microsoft Virtual Disk Service uses the CIM technology to list
information and manage LUNs. See the IBM System Storage DS Open Application
Programming Interface Reference, GC35-0516, for information about how to install and
configure VDS support.
The following sections present examples of VDS integration with advanced functions of the
DS8000 storage systems that became possible with the implementation of the DS CIM agent.
W in d o w s 2 0 0 3 W in d o w s 2 0 0 3
S e rv e r S e rv e r
W in d o w s V S S W in d o w s V S S
IB M A P I s u p p o r t fo r M ic r o s o f t I B M A P I s u p p o r t f o r M ic r o s o f t
V o lu m e S h a d o w C o p y S e r v ic e V o lu m e S h a d o w C o p y S e r v ic e
M a n a g e m e n t s e rv e r
D S C LI
C IM C lie n t E S S C LI
S M C S H M C
D S 6000 D S 8000 E S S
After the installation of these components —described in IBM System Storage DS Open
Application Programming Interface Reference, GC35-0516, you have to:
define a VSS_FREE volume group and virtual server
define a VSS_RESERVED volume group and virtual server
assign volumes to the VSS_FREE volume group.
The WWPN default for the VSS_FREE virtual server is 50000000000000; for the
VSS_RESERVED virtual server is 50000000000001. This disks are available for the server
like a pool of free available disks. If you want to have different pools of free disks you can
define your own WWPN for another pool; see Example 15-3.
Commands:
/h | /help | -? | /?
showcfg
listvols <all|free|vss|unassigned>
add <volumeID list> (separated by spaces)
rem <volumeID list> (separated by spaces)
Configuration:
set targetESS <5-digit ESS Id>
set user <CIMOM user name>
set password <CIMOM password>
set trace [0-7]
set trustpassword <trustpassword>
set truststore <truststore location>
set usingSSL <YES | NO>
set vssFreeInitiator <WWPN>
set vssReservedInitiator <WWPN>
set FlashCopyVer <1 | 2>
set cimomPort <PORTNUM>
set cimomHost <Hostname>
set namespace <Namespace>
With the ibmvssconfig.exe listvols command you can also verify what volumes are
available for VSS in the VSS_FREE pool; see Example 15-4.
Also disks that are unassigned in your disk subsystem can be assigned with the add
command to the VSS_FREE pool. In Example 15-5 on page 298 we verify the volumes
available for VSS.
Command
Command
Requestor Line
Line
Backup Interface
Interface
App Writers
Writers
Virtual Storage
Storage
Disk Mgt
Apps Service App
App
Volume
Shadow
Copy I/O Disk
Disk Mgt
Mgt
Service
DS8000/
DS8000/ IBM VSS Hardware IBM VDS
Provider Provider
DS6000
DS6000 Provider
FlashCopy
Target Source
Win2003 Win2003
Backup Production
Server Server
IBM has a GDS service offering to deliver solutions in this area —visit the IBM web site and
see the Services & Industry Solutions page for more information.
For more information see IBM System Storage DS8000 Series: Copy Services in Open
Environments, SG24-6788.
15.3 AIX
This section covers items specific to the IBM AIX operating system. It is not intended to
repeat the information that is contained in other publications. We focus on topics that are not
covered in the well-known literature or are important enough to be repeated here.
For complete information refer to IBM System Storage DS8000 Host Systems Attachment
Guide, SC26-7917.
Part Number.................00P4494
EC Level....................A
Serial Number...............1A31005059
Manufacturer................001A
Feature Code/Marketing ID...2765
FRU Number.................. 00P4495
Network Address.............10000000C93318D6
ROS Level and ID............02C03951
Device Specific.(Z0)........2002606D
Device Specific.(Z1)........00000000
Device Specific.(Z2)........00000000
Device Specific.(Z3)........03000909
Device Specific.(Z4)........FF401210
Device Specific.(Z5)........02C03951
Device Specific.(Z6)........06433951
Device Specific.(Z7)........07433951
Device Specific.(Z8)........20000000C93318D6
Device Specific.(Z9)........CS3.91A1
Device Specific.(ZA)........C1D3.91A1
Device Specific.(ZB)........C2D3.91A1
Device Specific.(YL)........U1.13-P1-I1/Q1
The # stands for the instance of each FC HBA you want to query.
MPIO and SDD cannot coexist on the same server. See following sections for detailed
discussion and considerations.
The following file sets should not be installed and must be removed:
devices.fcp.disk.ibm.mpio.rte
devices.sddpcm.52.rte, or devices.sddpcm.53.rte
For datapath query device the option -l is added to mark the non preferred paths in a
storage unit. This option can be used in addition to the existing datapath query device
commands. In Example 15-7, DS8000 disks are mixed with DS6000 devices. Because the
DS8000 does not implement the concept of preferred data path, you see non preferred paths
marked with an asterisk (*) only for the DS6000 volumes. On the DS8000 all paths are used.
Total Devices : 4
The datapath query portmap command shows the usage of the ports. In Example 15-8 you
see a mixed DS8000 and DS6000 disks configuration seen by the server. For the DS6000 the
datapath query portmap command uses capital letters for the preferred paths and lower case
letters for non preferred paths —this does not apply to the DS8000.
Note: 2105 devices' essid has 5 digits, while 1750/2107 device's essid has 7 digits.
Sometimes the lsvpcfg command helps you get an overview of your configuration. You can
easily count how many physical disks there are, with which serial number, and how many
paths. See Example 15-9.
There are also some other valuable features in SDD for AIX:
Enhanced SDD configuration methods and migration.
SDD has a feature in the configuration method to read the pvid from the physical disks and
convert the pvid from hdisks to vpaths during the SDD vpath configuration. With this
feature, you can skip the process of converting the pvid from hdisks to vpaths after
configuring SDD devices. Furthermore, SDD migration can skip the pvid conversion
process. This tremendously reduces the SDD migration time, especially with a large
number of SDD devices and LVM configuration environment.
Allow mixed volume groups with non-SDD devices in hd2vp, vp2hd, and dpovgfix.
Mixed volume group is supported by three SDD LVM conversion scripts: hd2vp, vp2hd,
and dpovgfix. These three SDD LVM conversion script files allow pvid conversion even if
the volume group consists of SDD supported devices and non-SDD supported devices.
Non-SDD supported devices allowed are IBM RDAC, EMC Powerpath, NEC MPO, and
Hitachi Dynamic Link Manager devices.
The base functionality of MPIO is limited. It provides an interface for vendor-specific Path
Control Modules (PCMs) which allow for implementation of advanced algorithms.
IBM provides a PCM for DS8000 that enhances MPIO with all the features of the original
SDD. It is called SDDPCM and is available from the SDD download site.
For basic information about MPIO see the online guide AIX 5L System Management
Concepts: Operating System and Devices from the AIX documentation site, at:
http://publib16.boulder.ibm.com/pseries/en_US/aixbman/admnconc/hotplug_mgmt.htm#mpioconcepts
The management of MPIO devices is described in the online guide System Management
Guide: Operating System and Devices for AIX 5L from the AIX documentation site, at:
http://publib16.boulder.ibm.com/pseries/en_US/aixbman/baseadmn/manage_mpio.htm
For information about SDDPCM commands, refer to the IBM System Storage Multipath
Subsystem Device Driver User’s Guide, SC30-4131. The SDDPCM Web site is located at:
http://www-1.ibm.com/servers/storage/support/software/sdd/index.html
Benefits of MPIO
There are some reasons to prefer MPIO with SDDPCM to traditional SDD:
Performance improvements due to direct integration with AIX
Better integration if different storage systems are attached
Easier administration through native AIX commands
DS8000 volumes to the operating system as MPIO manageable. Of course, you can’t have
SDD and MPIO/SDDPCM on a given server at the same time.
A point worth considering when deciding between SDD and MPIO is, that the IBM System
Storage SAN Volume Controller does not support MPIO at this moment. For updated
information refer to:
http://www.ibm.com/servers/storage/software/virtualization/svc/index.html
Other than with SDD, each disk will only be presented one time, and you can use normal AIX
commands; see Example 15-10. The DS8000 disk will only be seen one time as IBM MPIO
FC 2107.
Like SDD, MPIO with PCM supports the preferred path of DS6000 — in the DS8000 there are
no preferred paths. The algorithm of load leveling can be changed like SDD.
Example 15-11 shows a pcmpath query device command for a mixed environment, with two
DS8000s and one DS6000 disk.
All other commands are like with SDD, such as pcmpath query essmap or pcmpath query
portmap. In Example 15-12 you see these commands in a mixed environment with two
DS8000 disks and one DS6000 disk.
Example 15-12 MPIO pcmpath queries in a mixed DS8000 and DS6000 environment
root@san5198b:/ > pcmpath query essmap
Disk Path P Location adapter LUN SN Type Size LSS Vol Rank C/A S Connection port RaidMod
e
------- ----- - ---------- -------- -------- ------------ ---- --- --- ---- --- - ----------- ---- -------
-
hdisk2 path0 1p-20-02[FC] fscsi0 75065711100 IBM 2107-900 5.0 17 0 0000 17 Y R1-B1-H1-ZB 1 RAID5
hdisk2 path1 1p-20-02[FC] fscsi0 75065711100 IBM 2107-900 5.0 17 0 0000 17 Y R1-B2-H3-ZB 131 RAID5
hdisk2 path2 1p-20-02[FC] fscsi0 75065711100 IBM 2107-900 5.0 17 0 0000 17 Y R1-B3-H4-ZB 241 RAID5
hdisk2 path3 1p-20-02[FC] fscsi0 75065711100 IBM 2107-900 5.0 17 0 0000 17 Y R1-B4-H2-ZB 311 RAID5
hdisk3 path0 1p-20-02[FC] fscsi0 75065711101 IBM 2107-900 5.0 17 1 0000 17 Y R1-B1-H1-ZB 1 RAID5
hdisk3 path1 1p-20-02[FC] fscsi0 75065711101 IBM 2107-900 5.0 17 1 0000 17 Y R1-B2-H3-ZB 131 RAID5
hdisk3 path2 1p-20-02[FC] fscsi0 75065711101 IBM 2107-900 5.0 17 1 0000 17 Y R1-B3-H4-ZB 241 RAID5
hdisk3 path3 1p-20-02[FC] fscsi0 75065711101 IBM 2107-900 5.0 17 1 0000 17 Y R1-B4-H2-ZB 311 RAID5
hdisk4 path0 * 1p-20-02[FC] fscsi0 13AAGXA1101 IBM 1750-500 10.0 17 1 0000 07 Y R1-B1-H1-ZA 0 RAID5
hdisk4 path1 1p-20-02[FC] fscsi0 13AAGXA1101 IBM 1750-500 10.0 17 1 0000 07 Y R1-B2-H1-ZA 100 RAID5
hdisk4 path2 * 1p-28-02[FC] fscsi1 13AAGXA1101 IBM 1750-500 10.0 17 1 0000 07 Y R1-B1-H1-ZA 0 RAID5
hdisk4 path3 1p-28-02[FC] fscsi1 13AAGXA1101 IBM 1750-500 10.0 17 1 0000 07 Y R1-B2-H1-ZA 100 RAID5
Note: 2105 devices' essid has 5 digits, while 1750/2107 device's essid has 7 digits.
Note that the non-preferred path asterisk is only for the DS6000.
The chpath command is used to perform change operations on a specific path. It can either
change the operational status or tunable attributes associated with a path. It cannot perform
both types of operations in a single invocation.
The rmpath command un-configures or un-defines, or both, one or more paths to a target
device. It is not possible to un-configure (un-define) the last path to a target device using the
rmpath command. The only way to do this is to un-configure the device itself (for example, use
the rmdev command).
LVM striping
Striping is a technique for spreading the data in a logical volume across several physical disks
in such a way that all disks are used in parallel to access data on one logical volume. The
primary objective of striping is to increase the performance of a logical volume beyond that of
a single physical disk.
In the case of a DS8000, LVM striping can be used to distribute data across more than one
array (rank).
Refer to Chapter 10, “Performance” on page 167 for a more detailed discussion of methods to
optimize performance.
LVM mirroring
LVM has the capability to mirror logical volumes across several physical disks. This improves
availability, because in case a disk fails, there will be another disk with the same data. When
creating mirrored copies of logical volumes, make sure that the copies are indeed distributed
across separate disks.
With the introduction of SAN technology, LVM mirroring can even provide protection against a
site failure. Using long wave Fibre Channel connections, a mirror can be stretched up to a 10
km distance.
Synchronous I/O
Synchronous I/O occurs while you wait. An application’s processing cannot continue until the
I/O operation is complete. This is a very secure and traditional way to handle data. It ensures
consistency at all times, but can be a major performance inhibitor. It also doesn’t allow the
operating system to take full advantage of functions of modern storage devices, such as
queueing, command reordering, and so on.
Asynchronous I/O
Asynchronous I/O operations run in the background and do not block user applications. This
improves performance, because I/O and application processing run simultaneously. Many
applications, such as databases and file servers, take advantage of the ability to overlap
processing and I/O. They have to take measures to ensure data consistency, though. You can
configure, remove, and change asynchronous I/O for each device using the chdev command
or SMIT.
Tip: If the number of async I/O (AIO) requests is high, then the recommendation is to
increase maxservers to approximately the number of simultaneous I/Os there might be. In
most cases, it is better to leave the minservers parameter to the default value since the AIO
kernel extension will generate additional servers if needed. By looking at the CPU
utilization of the AIO servers, if the utilization is even across all of them, that means that
they’re all being used; you may want to try increasing their number in this case. Running
pstat -a will allow you to see the AIO servers by name, and running ps -k will show them
to you as the name kproc.
Direct I/O
An alternative I/O technique called Direct I/O bypasses the Virtual Memory Manager (VMM)
altogether and transfers data directly from the user’s buffer to the disk and vice versa. The
concept behind this is similar to raw I/O in the sense that they both bypass caching at the file
system level. This reduces CPU overhead and makes more memory available to the
database instance, which can make more efficient use of it for its own purposes.
Direct I/O is provided as a file system option in JFS2. It can be used either by mounting the
corresponding file system with the mount –o dio option, or by opening a file with the
O_DIRECT flag specified in the open() system call. When a file system is mounted with the
–o dio option, all files in the file system use Direct I/O by default.
Direct I/O benefits applications that have their own caching algorithms by eliminating the
overhead of copying data twice, first between the disk and the OS buffer cache, and then from
the buffer cache to the application’s memory.
For applications that benefit from the operating system cache, Direct I/O should not be used,
because all I/O operations would be synchronous. Direct I/O also bypasses the JFS2
read-ahead. Read-ahead can provide a significant performance boost for sequentially
accessed files.
Concurrent I/O
In 2003, IBM introduced a new file system feature called Concurrent I/O (CIO) for JFS2. It
includes all the advantages of Direct I/O and also relieves the serialization of write accesses.
It improves performance for many environments, particularly commercial relational
databases. In many cases, the database performance achieved using Concurrent I/O with
JFS2 is comparable to that obtained by using raw logical volumes.
A method for enabling the concurrent I/O mode is to use the mount -o cio option when
mounting a file system.
15.4 Linux
Linux is an open source UNIX-like kernel, originally created by Linus Torvalds. The term
Linux is often used to mean the whole operating system of GNU/Linux. The Linux kernel,
along with the tools and software needed to run an operating system, are maintained by a
loosely organized community of thousands of (mostly) volunteer programmers.
There are several organizations (distributors) that bundle the Linux kernel, tools, and
applications to form a distribution, a package that can be downloaded or purchased and
installed on a computer. Some of these distributions are commercial; others are not.
15.4.1 Support issues that distinguish Linux from other operating systems
Linux is different from the other, proprietary, operating systems in many ways:
There is no one person or organization that can be held responsible or called for support.
Depending on the target group, the distributions differ largely in the kind of support that is
available.
Linux is available for almost all computer architectures.
Linux is rapidly changing.
All these factors make it difficult to promise and provide generic support for Linux. As a
consequence, IBM has decided on a support strategy that limits the uncertainty and the
amount of testing.
IBM only supports the major Linux distributions that are targeted at enterprise clients:
RedHat Enterprise Linux
SUSE Linux Enterprise Server
These distributions have release cycles of about one year, are maintained for five years, and
require the user to sign a support contract with the distributor. They also have a schedule for
regular updates. These factors mitigate the issues listed previously. The limited number of
supported distributions also allows IBM to work closely with the vendors to ensure
interoperability and support. Details about the supported Linux distributions can be found in
the DS8000 Interoperability Matrix:
http://www-1.ibm.com/servers/storage/disk/DS8000/interop.html
There are exceptions to this strategy when the market demand justifies the test and support
effort.
It is not very detailed with respect to the configuration and installation of the FC HBA drivers.
IBM System z dedicates its own Web page to storage attachment via FCP:
http://www.ibm.com/servers/eserver/zseries/connectivity/ficon_resources.html
The System z connectivity support page lists all supported storage devices and SAN
components that can be attached to a System z server. There is an extra section for FCP
attachment:
http://www.ibm.com/servers/eserver/zseries/connectivity/#fcp
Another whitepaper, Linux on IBM eServer pSeries SAN - Overview for Customers, describes
in detail how to attach SAN storage (ESS 2105 and DS4000(former FAStT)) to a System p
server running Linux:
http://www.ibm.com/servers/eserver/pseries/linux/whitepapers/linux_san.pdf
Most of the information provided in these publications is valid for DS8000 attachment,
although much of it was originally written for the ESS 2105.
8 /dev/sda /dev/sdp
65 /dev/sdq /dev/sdaf
66 /dev/sdag /dev/sdav
71 /dev/sddi /dev/sddx
Major number First special device file Last special device file
Each SCSI device can have up to 15 partitions, which are represented by the special device
files /dev/sda1™, /dev/sda2, and so on. The mapping of partitions to special device files and
major and minor numbers is shown in Table 15-2.
...
...
...
... ...
Refer to the man page of the mknod command for more details. Example 15-14 on page 311
shows the creation of special device files for the seventeenth SCSI disk and its first three
partitions.
Example 15-14 Create new special device files for SCSI disks
mknod /dev/sdq b 65 0
mknod /dev/sdq1 b 65 1
mknod /dev/sdq2 b 65 2
mknod /dev/sdq3 b 65 3
After creating the device files you may have to change their owner, group, and file permission
settings to be able to use them. Often, the easiest way to do this is by duplicating the settings
of existing device files, as shown in Example 15-15. Be aware that after this sequence of
commands, all special device files for SCSI disks have the same permissions. If an
application requires different settings for certain disks, you have to correct them afterwards.
The IBM multipathing solution for DS8000 attachment to Linux on Intel IA-32 and IA-64
architectures, IBM System p, and System i is the IBM Subsystem Device Driver (SDD). SDD
for Linux is available in the Linux RPM package format for all supported distributions from the
SDD download site. It is proprietary and binary only. It only works with certain kernel versions
with which it was tested. The readme file on the SDD for Linux download page contains a list
of the supported kernels.
The version of the Linux Logical Volume Manager that comes with all current Linux
distributions does not support its physical volumes being placed on SDD vpath devices.
Non-root users can now open a vpath device. Before, only root users would have this
privilege, but with the new capabilities in the OS, non-root users can do the same.
Note: SDD is not available for Linux on System z. SUSE Linux Enterprise Server 8 for
System z comes with built-in multipathing provided by a patched Logical Volume Manager.
Today there is no multipathing support for Redhat Enterprise Linux for System z.
Important: The latest update to the SUSE Linux Enterprise Server 8, Service Pack 3 uses
a more dynamic method of assigning major numbers and allows the attachment of up to
2304 SCSI devices.
Multiple LUN support can be added with an option to the SCSI midlayer Kernel module
scsi_mod. To have multiple LUN support added permanently at boot time of the system, add
the following line to the file /etc/modules.conf: options scsi_mod max_scsi_luns=128.
After saving the file, rebuild the module dependencies by running: depmod -a
Now you have to rebuild the Initial RAM Disk, using the command:
mkinitrd <initrd-image> <kernel-version>
Issue mkinitrd -h for more help information. A reboot is required to make the changes
effective.
To prevent the FC HBA driver from being loaded before the driver for the internal SCSI HBA
you have to change the /etc/modules.conf file:
Locate the lines containing scsi_hostadapterx entries, where x is a number.
Reorder these lines: First come the lines containing the name of the internal HBA driver
module, then the ones with the FC HBA module entry.
Renumber lines: No number for the first entry, 1 for the second, 2 for the third, and so on.
After saving the file, rebuild the module dependencies by running: depmod -a
Now you have to rebuild the Initial RAM Disk, using the command:
mkinitrd <initrd-image> <kernel-version>
Issue mkinitrd -h for more help information. If you reboot now, the SCSI and FC HBA drivers
will be loaded in the correct order.
Example 15-16 shows how the /etc/modules.conf file should look with two Adaptec SCSI
controllers and two QLogic 2340 FC HBAs installed. It also contains the line that enables
multiple LUN support. Note that the module names will be different with different SCSI and
Fibre Channel adapters.
A Linux system can recognize newly attached LUNs without unloading the FC HBA driver.
The procedure slightly differs depending on the installed FC HBAs.
This script is not part of the regular device driver package and must be downloaded
separately: http://www.emulex.com/ts/downloads/linuxfc/rel/201g/force_lpfc_scan.sh
In both cases the command must be issued for each installed HBA, with the
<adapter-instance> being the SCSI instance number of the HBA.
After the FC HBAs rescan the fabric, you can make the new devices available to the system
with the command echo "scsi add-single-device s c t l" > /proc/scsi/scsi.
The new volumes are added after the already existing ones. The following examples illustrate
this. Example 15-17 shows the original disk assignment as it existed since the last system
start.
Example 15-18 shows the SCSI disk assignment after one more DS8000 volume is added.
Example 15-18 SCSi disks after dynamic addition of another DS8000 volume
/dev/sda - internal SCSI disk
/dev/sdb - 1st DS8000 volume, seen by HBA 0
/dev/sdc - 2nd DS8000 volume, seen by HBA 0
/dev/sdd - 1st DS8000 volume, seen by HBA 1
/dev/sde - 2nd DS8000 volume, seen by HBA 1
/dev/sdf - new DS8000 volume, seen by HBA 0
/dev/sdg - new DS8000 volume, seen by HBA 1
The mapping of special device files is now different than it would have been if all three
DS8000 volumes had already been present when the HBA driver was loaded. In other words,
if the system is now restarted, the device ordering will change to what is shown in
Example 15-19.
Example 15-19 SCSi disks after dynamic addition of another DS8000 volume and reboot
/dev/sda - internal SCSI disk
/dev/sdb - 1st DS8000 volume, seen by HBA 0
/dev/sdc - 2nd DS8000 volume, seen by HBA 0
/dev/sdd - new DS8000 volume, seen by HBA 0
/dev/sde - 1st DS8000 volume, seen by HBA 1
/dev/sdf - 2nd DS8000 volume, seen by HBA 1
/dev/sdg - new DS8000 volume, seen by HBA 1
When assigning volumes to a Linux host with QLogic FC HBAs, make sure LUNs start at 0
and are in consecutive order. Otherwise the LUNs after a gap will not be discovered by the
host. Gaps in the sequence can occur when you assign volumes to a Linux host that are
already assigned to another server.
The Emulex HBA driver behaves differently: It always scans all LUNs up to 127.
The file /proc/scsi/scsi contains a list of all attached SCSI devices, including disk, tapes,
processors, and so on. Example 15-20 shows a sample /proc/scsi/scsi file.
There is also an entry in /proc for each HBA, with driver and firmware levels, error counters,
and information about the attached devices. Example 15-21 shows the condensed content of
the entry for a QLogic Fibre Channel HBA.
By default SLES 8 provides sg device files for up to 16 SCSI devices (/dev/sg0 through
/dev/sg15). Additional sg device files can be created using the command mknod. After creating
new sg devices you should change their group setting from root to disk. Example 15-22
shows the creation of /dev/sg16, which would be the first one to create.
Example 15-22 Creation of new device files for generic SCSI devices
mknod /dev/sg16 c 21 16
chgrp disk /dev/sg16
sginfo prints SCSI inquiry and mode page data; it also allows manipulating the mode pages.
15.5 OpenVMS
DS8000 supports FC attachment of OpenVMS Alpha systems with operating system
Version 7.3 or later. For details regarding operating system versions and HBA types, see the
DS8000 Interoperability Matrix, available at:
http://www.ibm.com/servers/storage/disk/ds8000/interop.html
The support includes clustering and multiple paths (exploiting the OpenVMS built-in
multipathing). Boot support is available via Request for Price Quotations (RPQ).
Instead of writing a special OpenVMS driver, it had been decided to handle this in the
DS8000 host adapter microcode. As a result, it became a general rule not to share DS8000
FC ports between OpenVMS and non-OpenVMS hosts.
Important: The DS8000 FC ports used by OpenVMS hosts must not be accessed by any
other operating system, not even accidentally. The OpenVMS hosts have to be defined for
access to these ports only, and it must be ensured that no foreign HBA (without definition
as an OpenVMS host) is seen by these ports. Conversely, an OpenVMS host must have
access only to the DS8000 ports configured for OpenVMS compatibility.
You must dedicate storage ports for only the OpenVMS host type. Multiple OpenVMS
systems can access the same port. Appropriate zoning must be enforced from the beginning.
Wrong access to storage ports used by OpenVMS hosts may clear the OpenVMS-specific
settings for these ports. This might remain undetected for a long time—until some failure
happens, and by then I/Os might be lost. It is worth mentioning that OpenVMS is the only
platform with such a restriction (usually, different open systems platforms can share the same
DS8000 FC adapters).
The restrictions listed in this section apply only if your DS8000 licensed machine code has a
version before 5.0.4. After these versions, the restrictions are removed. Note that you can
display the versions of the DS CLI, the DS Storage Manager, and the licensed machine code
by using the DS CLI command ver -l.
OpenVMS does not identify a Fibre Channel disk by its path or SCSI target/LUN like other
operating systems. It relies on the UDID. Although OpenVMS uses the WWID to control all FC
paths to a disk, a Fibre Channel disk that does not provide this additional UDID cannot be
recognized by the operating system.
In the DS8000, the volume nickname acts as the UDID for OpenVMS hosts. If the character
string of the volume nickname evaluates to an integer in the range 0–32767, then this integer
is replied as the answer when an OpenVMS host asks for the UDID.
The DS CLI command chfbvol -name 21 1001 assigns the OpenVMS UDID 21 to the
DS8000 volume 1001 (LSS 10, volume 01). Thus the DS8000 volume 1001 will appear as an
OpenVMS device with the name $1$DGA21 or $1$GGA21. The DS CLI command lshostvol
shows the DS8000 volumes with their corresponding OpenVMS device names.
The DS management utilities do not enforce UDID rules. They accept incorrect values that
are not valid for OpenVMS. It is possible to assign the same UDID value to multiple DS8000
volumes. However, because the UDID is in fact the device ID seen by the operating system,
several consistency rules have to be fulfilled. These rules are described in detail in the
OpenVMS operating system documentation; see HP Guidelines for OpenVMS Cluster
Configurations at http://h71000.www7.hp.com/doc/72final/6318/6318pro.html:
Every FC volume must have a UDID that is unique throughout the OpenVMS cluster that
accesses the volume. The same UDID may be used in a different cluster or for a different
stand-alone host.
If the volume is planned for MSCP serving, then the UDID range is limited to 0–9999 (by
operating system restrictions in the MSCP code).
OpenVMS system administrators tend to use elaborate schemes for assigning UDIDs, coding
several hints about physical configuration into this logical ID, for instance, odd/even values or
reserved ranges to distinguish between multiple data centers, storage systems, or disk
groups. Thus they must be able to provide these numbers without additional restrictions
imposed by the storage system. In the DS8000, UDID is implemented with full flexibility, which
leaves the responsibility about restrictions to the user.
In Example 15-23, we configured a DS8000 volume with the UDID 8275 for OpenVMS
attachment. This gives the OpenVMS Fibre Channel disk device $1$DGA8275. You see the
output from the OpenVMS command show device/full $1$DGA8275. The OpenVMS host
has two Fibre Channel HBAs with names PGA0 and PGB0. Because each HBA accesses
two DS8000 ports, we have four I/O paths.
The DS CLI command lshostvol displays the mapping of DS8000 volumes to host system
device names. More details regarding this command can be found in the IBM System Storage
DS8000: Command-Line Interface User´s Guide, SC26-7916.
Because the OpenVMS FC driver has been written with StorageWorks controllers in mind,
OpenVMS always considers LUN 0 as CCL, never presenting this LUN as a disk device. On
HP StorageWorks HSG and HSV controllers, you cannot assign LUN 0 to a volume.
The DS8000 assigns LUN numbers per host using the lowest available number. The first
volume that is assigned to a host becomes this host’s LUN 0; the next volume is LUN 1, and
so on.
Because OpenVMS considers LUN 0 as CCL, the first DS8000 volume assigned to the host
cannot be used even when a correct UDID has been defined. So we recommend creating the
first OpenVMS volume with a minimum size as a dummy volume for use as the CCL. Multiple
OpenVMS hosts, even in different clusters, that access the same storage system can share
the same volume as LUN 0, because there will be no other activity to this volume. In large
configurations with more than 256 volumes per OpenVMS host or cluster, it might be
necessary to introduce another dummy volume (when LUN numbering starts again with 0).
Defining a UDID for the CCL is not required by the OpenVMS operating system. OpenVMS
documentation suggests that you always define a unique UDID since this identifier causes the
creation of a CCL device visible for the OpenVMS command show device or other tools.
Although an OpenVMS host cannot use the LUN for any other purpose, you can display the
multiple paths to the storage device, and diagnose failed paths. Fibre Channel CCL devices
have the OpenVMS device type GG.
In Example 15-24, the DS8000 volume with volume ID 100E is configured as an OpenVMS
device with UDID 9998. Because this was the first volume in the volume group, it became
LUN 0 and thus the CCL. Please note that the volume WWID, as displayed by the show
device/full command, contains the DS8000 World-Wide Node ID (6005-0763-03FF-C324)
and the DS8000 volume number (100E).
The DS CLI command chvolgrp provides the flag -lun which can be used to control which
volume becomes LUN 0.
The OpenVMS shadow driver has been designed for disks according to DEC’s Digital
Storage Architecture (DSA). This architecture, forward-looking in the 1980s, includes some
requirements that are handled by today’s SCSI/FC devices with other approaches. Two such
things are the forced error indicator and the atomic revector operation for bad-block
replacement.
When a DSA controller detects an unrecoverable media error, a spare block is revectored to
this logical block number, and the contents of the block are marked with a forced error. This
causes subsequent read operations to fail, which is the signal to the shadow driver to execute
a repair operation using data from another copy.
However, there is no forced error indicator in the SCSI architecture, and the revector
operation is nonatomic. As a substitute, the OpenVMS shadow driver exploits the SCSI
commands READ LONG (READL) and WRITE LONG (WRITEL), optionally supported by
some SCSI devices. These I/O functions allow data blocks to be read and written together
with their disk device error correction code (ECC). If the SCSI device supports
READL/WRITEL, OpenVMS shadowing emulates the DSA forced error with an intentionally
incorrect ECC. For details see Scott H. Davis, Design of VMS Volume Shadowing Phase II —
Host-based Shadowing, Digital Technical Journal Vol. 3 No. 3, Summer 1991, archived at:
http://research.compaq.com/wrl/DECarchives/DTJ/DTJ301/DTJ301SC.TXT
The DS8000 provides volumes as SCSI-3 devices and thus does not implement a forced
error indicator. It also does not support the READL and WRITEL command set for data
integrity reasons.
Usually the OpenVMS SCSI Port Driver recognizes if a device supports READL/WRITEL, and
the driver sets the NOFE (no forced error) bit in the Unit Control Block. You can verify this
setting with the SDA utility: After starting the utility with the analyze/system command, enter
the show device command at the SDA prompt. Then the NOFE flag should be shown in the
device’s characteristics.
15.6 VMware
The DS8000 currently supports VMware high-end virtualization product, the ESX Server,
starting with Version 2.5. The supported guest operating systems are Windows 2000,
Windows Server 2003, SUSE Linux SLES 8, and Red Hat Enterprise Linux 2.1 and 3.0. This
information is likely to change, so check the Interoperability matrix for complete, up-to-date
information: http://www.ibm.com/servers/storage/disk/ds8000/interop.html
A great deal of useful information is available in the IBM System Storage DS8000 Host
Systems Attachment Guide, SC26-7917. This section is not intended to duplicate that
publication, but rather it provides more information about optimizing your VMware
environment as well as a step-by-step guide to setting up ESX Server with the DS8000.
VMware other products, such as GSX Server and Workstation, are not intended for the
datacenter-class environments where the DS8000 typically is used. Certain other products,
such as VMotion and VirtualCenter, may be supported on a case-by-case basis. Before using
the techniques describe below, check with IBM and the latest Interoperability matrix for the
supportability of these techniques.
ESX Server 2.5 also introduced support for booting directly from the SAN. This feature is not
yet fully supported with the DS8000, though support may be available via RPQ. See the
DS8000 Interoperability Matrix for the latest information on support for SAN boot.
In Figure 15-6, VM1, VM2, and VM3 use .dsk files stored on a SAN disk, while VM4 directly
uses a raw SAN disk. VMs can also use the physical server’s local storage (as the Console
OS does), but these disks tend to be not as fast or reliable as SAN disks. Both the virtual
machine .dsk files and the raw disks represent what is seen as physical disks by the guest
OS.
As with other operating systems, you should have multiple paths from your server to the
DS8000 to improve availability and reliability. Normally, the LUNs would show up as multiple
separate devices, but VMware contains native multipathing software that automatically
conceals the redundant paths. Therefore, multipathing software may not be needed on your
guest operating systems.
As with other operating systems, you should also use persistent binding. See the IBM System
Storage DS8000 Host Systems Attachment Guide, SC26-7917, on why persistent binding is
important and how to configure it for VMware.
After the LUNs are assigned properly, you will be able to see them from the VMware
administration console, under Options → Storage Management → Failover Paths. You
may have to tell VMware to refresh its disks by selecting Rescan SAN in the upper right-hand
corner. Figure 15-7 shows the LUNs assigned to the ESX Server and the paths to each of the
LUNs.
Note: The virtual machine should be powered off before adding hardware.
To select the appropriate type, select the appropriate VM, choose the Hardware tab and then
Add Device.
Option 1: Formatting these disks with the VMFS: This option maximizes the virtualization
features that are possible, and allows the guest operating system to use the special
features of VMFS volumes. However, this mode has the most overhead of the three
options.
– After clicking Add Device, choose Hard Disk and then select the Blank type.
– Select the options for the new hard disk that are appropriate for your environment.
Option 2: Passing the disk through to the guest OS as a raw disk in physical compatibility
mode: No further virtualization occurs; the OS will write its own file system onto that disk
directly, just as it would in a standalone environment, without an underlying VMFS
structure. I/Os pass through the virtualization layer with minimal modification. This option
requires the least overhead.
– After clicking Add Device, choose Hard Disk then select the System LUN/Disk type.
– On the next panel, choose the compatibility mode of Physical.
– Select the options for the new hard disk that are appropriate for your environment.
Option 3: Passing the disk through to the guest OS as a raw disk in virtual compatibility
mode: This mode allows the VM to take advantage of disk modes and other features,
including redo logs. Please see the VMware documentation on the four different disk
modes: persistent, nonpersistent, undoable, and append.
– After clicking Add Device, choose Hard Disk then select the System LUN/Disk type.
– On the next screen, choose the compatibility mode of Virtual.
– Select the options for the new hard disk that are appropriate for your environment.
In Figure 15-8 on page 324, Virtual Disk (SCSI 0:1) is a VMware virtual disk, while Virtual
Disk (SCSI 0:2) is a physical SAN LUN and Virtual Disk (SCSI 0:3) is a virtual SAN LUN. This
VM also contains a local virtual disk, Virtual Disk (SCSI 0:0), which has the guest operating
system installed.
After powering up the server, notice how in Figure 15-9, the raw disk in Physical compatibility
mode shows up as an IBM 2107900 device, while the other three disks (the local disk, the
SAN disk formatted with VMFS, and the SAN disk in virtual compatibility mode) all show up
as VMware virtual disks.
Now the disks can be formatted and used as with any regular disks. System LUNs in physical
compatibility mode have the additional advantage that SCSI commands pass down to the
hardware with minimal modifications. As a result, system administrators can use the DS CLI
command lshostvol to map the virtual machine disks to DS8000 disks.
A great deal of useful information is available in the IBM System Storage DS8000 Host
Systems Attachment Guide, SC26-7917. This section is not intended to duplicate that
publication, but rather it provides more information about optimizing your Sun Solaris
environment as well as a step-by-step guide on using Solaris with the DS8000.
In many cases, you will also be able to use the prtconf command to list the WWPNs, as
shown in Example 15-25.
As with other operating systems, you should use persistent binding with Solaris and DS8000.
If you do not use persistent binding, it is possible that Solaris will assign a different SCSI
device identifier (SCSI ID) than the one it had been using previously. This can happen if a new
device is added to the SAN, for example. In this case, you will have to re-configure your
applications or your operating system.
The methods of enabling persistent binding differ depending on your host bus adapter. The
IBM System Storage DS8000 Host Systems Attachment Guide, SC26-7917, contains the
recommended HBA settings for each supported type.
First, IBM provides the System Storage Multipath Subsystem Device Driver (SDD) as a part
of the DS8000 at no extra charge. Next, Sun Solaris contains native multipathing software
called the StorEdge Traffic Manager Software (STMS). STMS is commonly known as MPxIO
(multiplexed I/O) in the industry, and the remainder of this section will refer to this technology
as MPxIO. Finally, IBM supports VERITAS Volume Manager (VxVM) Dynamic Multipathing
(DMP), a part of the VERITAS Storage Foundation suite.
The multipathing technology that you should use depends a great deal on your operating
environment and, of course, your business requirements. There are some limitations
depending on your operating system version, your host bus adapters, and whether you use
clustering. Details are available in the IBM System Storage DS8000 Host Systems
Attachment Guide, SC26-7917.
One difference between the multipathing technologies is in whether they suppress the
redundant paths to the storage. MPxIO and DMP both suppress all paths to the storage
except for one, and the device appears to the application as a single-path device. SDD, on the
other hand, allows the original paths to be seen, but creates its own virtual device (called a
vpath) for applications to use.
If you assign LUNs to your server before you install multipathing software, you can see each
of the LUNs show up as two or more devices, depending on how many paths you have. In
Example 15-26, the iostat -nE command shows that the volume 75207814206 appears
twice—once as c2t1d1 on the first HBA, and once as c3t1d1 on the second HBA.
After the SDD software is installed, you can see that the paths have been grouped into virtual
vpath devices. Example 15-27 shows the output of the showvpath command.
For each device, the operating system creates a node in the /dev/dsk and /dev/rdsk
directories. After SDD is installed, you can see these new vpaths by listing the contents of
those directories. Note that with SDD, the old paths are not suppressed. Instead, new vpath
devices show up as /dev/rdsk/vpath1a, for example. When creating your volumes and file
systems, be sure to use the vpath device instead of the original device.
SDD also offers some parameters that you can tune for your environment. Specifically, SDD
offers three different load balancing schemes:
Failover
– No load balancing
– Second path is used only if the preferred path fails
Round robin
– Paths to use are chosen at random (but different paths than most recent I/O)
– If only two paths, then they alternate
Load balancing
– Path chosen based on estimated path load
– Default policy
The policy can be set through the use of the datapath set device policy command.
Before you enable MPxIO, you will want to configure your host bus adapters. Issue the cfgadm
-la command to see the current state of your adapters. Example 15-28 on page 328 shows
two adapters, c3 and c4, of type fc.
Note how the command reports that both adapters are un-configured. To configure the
adapters, issue the cfgadm -c configure cX (where X is the adapter number, 3 and 4 in this
case). Now, both adapters should show up as configured.
To configure your MPxIO, you will need to first enable it by editing the
/kernel/drv/scsi_vhci.conf file. For Solaris 10, you will need to edit the
/kernel/drv/fp.conf file instead. Find and change the mpxio-disable parameter to no:
mpxio-disable="no";
Next, add the following stanza to supply the vendor identification (VID) and product
identification (PID) information to MPxIO in the /kernel/drv/scsi_vhci.conf file:
device-type-scsi-options-list =
"IBM 2107900", "symmetric-option";
symmetric-option = 0x1000000;
The vendor string must be exactly 8 bytes, so you must type IBM followed by 5 spaces. Finally,
the system must be rebooted. After the reboot, MPxIO will be ready to be used.
For more information about MPxIO, including all the MPxIO commands and tuning
parameters, see the Sun Web site: http://www.sun.com/storage/software/.
During device discovery, the vxconfigd daemon compares the serial numbers of the different
devices. If two devices have the same serial number, then they are the same LUN, and DMP
will combine the paths. Listing the contents of the /dev/vx/rdmp directory will show only one
set of devices.
The vxdisk path command also demonstrates DMP’s path suppression capabilities. In
Example 15-29, devices c6t1d0s2 and c7t2d0s2 are combined into c6t1d0s2.
Now, you create volumes using the device name listed under the DANAME column. In
Figure 15-10, a volume is created using four disks, even though there are actually eight paths.
As with other multipathing software, DMP provides a number of parameters that you can tune
in order to maximize the performance and availability in your environment. For example, it is
possible to set a load balancing policy to dictate how the I/O should be shared between the
different paths. It is also possible to select which paths get used in which order in case of a
failure.
Complete details about the features and capabilities of DMP can be found on the VERITAS
Web site: http://www.veritas.com.
15.8 HP-UX
The DS8000 attachment is supported with HP-UX Version 11i or later. For providing a fault
tolerant connection to the DS8000, the HP Multipathing software PVLINKS or the IBM
Multipath Subsystem Device Driver (SDD) are supported.
For preparing the host to attach the DS8000, refer to the Infocenter at:
http://publib.boulder.ibm.com/infocenter/ds8000ic/index.jsp?topic=/com.ibm.storage
.ess.console.base.help.doc/f2c_attchnghpux_1tlxvv.html and select Configuring →
Attaching Hosts → Hewlett-Packard Server (HP-UX) host attachment.
For Installation of the SDD refer to IBM System Storage Multipath Subsystem Device Driver
User’s Guide, SC30-4131. The User’s Guide is available at the download page for each
individual SDD Operating System Version at:
http://www.ibm.com/support/dlsearch.wss?rs=540&tc=ST52G7&dc=D430.
SDD is a multipathing software with policy-based load balancing on all available paths to the
DS8000. For automation purposes of the Copy Services, storage management, and storage
allocation, the DS CLI should be installed on the host.
For the latest version of SDD and the DS CLI on your host, you can either use the version that
is delivered with the DS8000 Microcode bundle or you can download the latest available SDD
version from: http://www.ibm.com/support/dlsearch.wss?rs=540&tc=ST52G7&dc=D430.
The latest available ISO-image for the DS CLI CD can be downloaded from:
ftp://ftp.software.ibm.com/storage/ds8000/updates/CLI/.
After the volumes on the DS8000 have been configured, install the SDD, then connect your
host to the fabric. Once the host is configured to the fabric, just discover the devices using the
ioscan command. Example 15-30 on page 331 does show that the DS8000 devices have
been discovered successfully, but the devices cannot be used, as no special device file is
available.
To create the missing special device file, there are two options; the first one is a reboot of the
host, which is disruptive. The alternative to the reboot is to run the command insf -eC disk,
which will reinstall the special device files for all devices of the Class disk. After creating the
special device files the ioscan output should look like Example 15-31.
Once the volumes are visible like in Example 15-31, volume groups (VGs), logical volumes,
and file systems can be created. If you have multiple paths to your DS8000, note that you
have to use as the /dev/dsk/vpathX device for creating the VGs.
15.8.4 Multipathing
The IBM Multipath Subsystem Device Driver (SDD) is a multipathing software that is capable
of policy-based load balancing on all available paths to the DS8000. The load balancing is a
major advantage.
PVLINKS is multipathing software that is built into the LVM of HP-UX. This software is only
performing a path failover to the alternate path once the primary path is not available any
more. In a poorly designed fail-over configuration, a performance bottleneck can be
produced. The bottleneck can arise, if all devices will only accessed via one adapter, but the
additional adapters are idle.
Precautions have to be taken, as from now on, the /dev/dsk/vpathX devices have to be
chosen to create VGs.
If you have installed the SDD on an existing machine and you want to migrate your devices to
become vpath devices, use the command hd2vp, which will convert your volume group to
access the vpath devices instead of the /dev/dsk/cXtXdX devices.
Example 15-32 shows the output of the DS CLI command lshostvol. This command is an
easy way of displaying the relationship between disk device files (paths to the DS8000), the
configured DS8000 LUN serial number, and the assigned vpath device.
For the support of MC/ServiceGuard with SDD, refer to the latest version of the DS8000
Interoperability Matrix.
SDD troubleshooting
When all DS8000 volumes are visible after claiming them with ioscan, but are not configured
by the SDD, you can run the command cfgvpath -r to perform a dynamic re-configuration of
all SDD devices.
Example 15-33 Sample syslog.log entries for the SDD link failure events
Nov 10 17:49:27 dwarf vmunix: WARNING: VPATH_EVENT: device = vpath8 path = 0 offline
Nov 10 17:49:27 dwarf vmunix: WARNING: VPATH_EVENT: device = vpath9 path = 0 offline
Nov 10 17:49:27 dwarf vmunix: WARNING: VPATH_EVENT: device = vpath10 path = 0 offline
Nov 10 17:50:15 dwarf vmunix: WARNING: VPATH_EVENT: device = vpath11 path = 0 offline
.....
Nov 10 17:56:12 dwarf vmunix: NOTICE: VPATH_EVENT: device = vpath9 path = 0 online
Nov 10 17:56:12 dwarf vmunix: NOTICE: VPATH_EVENT: device = vpath8 path = 0 online
Nov 10 17:56:12 dwarf vmunix: NOTICE: VPATH_EVENT: device = vpath10 path = 0 online
Nov 10 17:56:12 dwarf vmunix: NOTICE: VPATH_EVENT: device = vpath11 path = 0 online
16
ESCON
For optimum availability, make ESCON host adapters available through all I/O enclosures. For
good performance, and depending on your workload characteristics, use at least eight
ESCON ports installed on four ESCON host adapters in the storage unit.
Note: When using ESCON channels only volumes in address group 0 can be accessed.
For this reason, if you will have a mix of CKD and FB volumes in the storage image, you
may want to reserve the first 16 LSSs (00-0F) for the ESCON accessed CKD volumes.
FICON
You also need to check for dependencies in the host hardware driver level and the supported
feature codes. Your IBM service representative can help you determine your current hardware
driver level on your mainframe processor complex. Examples of limited host server feature
support are (FC 3319) FICON Express2 LX and (FC 3320) FICON Express2 SX, which are
available only for the z890 and z990 host server models.
Nevertheless, check the most recent edition of the DS8000 Interoperability Matrix to see the
list of supported operating systems at:
http://www-03.ibm.com/servers/storage/disk/ds8000/interop.html
Important: Always review the latest edition of the Interoperability Matrix and the
Preventive Service Planning (PSP) bucket of the 2107 for software updates.
The PSP information can be found on the Resource Link Web site at:
http://www-1.ibm.com/servers/resourcelink/svc03100.nsf?OpenDatabase
Many of these program enhancements are initially available as APARs and PTFs for the
current releases of z/OS, and are later integrated into the following releases of z/OS. For this
reason we recommend you review the DS8000 PSP bucket for your current release of z/OS.
Scalability support
The IOS recovery was designed to support a small number of devices per control unit —and
a unit check was presented on all devices at failover. This did not scale well with a DS8000
that has the capability to scale up to 65,280 devices. Under these circumstances, we may
have had CPU or spin lock contention, or exhausted storage below the 16M line at device
failover, or both.
Starting with z/OS 1.4 and higher with the DS8000 software support, the IOS recovery has
been improved by consolidating unit checks at an LSS level instead of each disconnected
device. This consolidation shortens the recovery time as a result of I/O errors. This
enhancement is particularly important since the DS8000 has a much higher number of
devices compared to the predecessors (ESS IBM 2105). In the IBM 2105, we had 4K devices
and in the DS8000 we have up to 65,280 devices in a storage facility.
Benefits
With the enhanced scalability support, the following benefits are possible:
Common storage (CSA) usage (above and below the 16M line) is reduced.
IOS large block pool for error recovery processing and attention, and state change
interrupt processing, is located above the 16M line, thus reducing storage demand below
the 16M line.
Unit control blocks (UCB) are pinned and event notification facility (ENF) signalling during
channel path recovery.
These scalability enhancements provide additional performance improvements by:
– Bypassing dynamic pathing validation in channel recovery for reduced recovery I/Os.
– Reducing elapsed time, by reducing the wait time in channel path recovery.
remain within the 64K limit. But as today’s storage facilities tend to expand to even larger
capacities, we are approaching the 64K limitation at a very fast rate. This leaves no choice but
to plan for even larger volumes sizes.
Support has been enhanced to expand volumes to 65,520 cylinders, using existing 16 bit
cylinder addressing. This is often referred to as 64K cylinder volumes. Components and
products such as DADSM/CVAF, DFSMSdss, ICKDSF, and DFSORT, previously shipped with
32,760 cylinders, now also support 65,520 cylinders. Check point restart processing now
supports a checkpoint data set that resides partially or wholly above the 32,760 cylinder
boundary.
With the new LVS volumes, the VTOC has the potential to grow very large. Callers such as
DFSMSdss have to read the entire VTOC to find the last allocated DSCB —in cases where
the VTOC is very large, performance degradation could be experienced. A new interface is
implemented to return the high allocated DSCB on volumes initialized with an INDEX VTOC.
DFSMSdss uses this interface to limit VTOC searches and improve performance. The VTOC
has to be within the first 64K-1 tracks, while the INDEX can be anywhere on the volume.
Control unit single point of failure information was specified in a table and had to be updated
for each new control unit. Instead, with the present enhancement, we can use the Read
Availability Mask (PSF/RSD) command to retrieve the information from the control unit. By
doing this, there is no need to maintain a table for this information.
The channel subsystem and z/OS were enhanced to retry I/O over an alternate channel path.
This circumvents IPL failures that were due to the selection of the same faulty path to read
from the SYSRES device.
Both storage images of 9A2 or 9B2 LPAR capable machine must be defined either as 2107 or
2105. Do not define one image as UNIT=2107 and the other as UNIT=2105.
Starting with z9-109 processors, users can define an additional subchannel set with ID 1 (SS
1) on top of the existing subchannel set (SS 0) in a channel subsystem. With this additional
subchannel set, you can configure more than 2*63K devices for a channel subsystem. With
z/OS V1R7, you can define Parallel Access Volume (PAV) alias devices (device types 3380A,
3390A) of the 2105, 2107 and 1750 DASD control units to SS 1. Device numbers may be
duplicated across channel subsystems and subchannel sets.
Performance statistics
Two new sets of performance statistics that are reported by the DS8000 were introduced.
Since a logical volume is no longer allocated on a single RAID rank with a single RAID type or
single device adapter pair, the performance data is now provided with a set of rank
performance statistics and extent pool statistics. The RAID RANK reports are no longer
reported by RMF and IDCAMS LISTDATA batch reports —RMF and IDCAMS LISTDATA is
enhanced to report the logical volume statistics that are provided on the DS8000.
These reports consist of back-end counters that capture the activity between the cache and
the ranks in the DS8000 for each individual logical volume. These rank and extent pool
statistics are disk-system-wide instead of volume-wide only.
Also, new response and transfer statistics are available with the Postprocessor Cache Activity
report generated from SMF record 74 subtype 5. These statistics are provided at the
subsystem level in the Cache Subsystem Activity report and at the volume level in the Cache
Device Activity report. In detail, RMF provides the average response time and byte transfer
rate per read and write requests. These statistics are shown for the I/O activity (called host
adapter activity) and transfer activity from hard disk to cache and vice-versa (called disk
activity).
New reports have been designed for reporting FICON channel utilization. RMF also provides
support for remote mirror and copy link utilization statistics. This support is delivered by APAR
OA04877 —PTFs are available for z/OS V1R4.
Note: RMF cache reporting and the results of a LISTDATA STATUS command report a
cache size that is half the actual size. This is because the information returned represents
only the cluster to which the logical control unit is attached. Each LSS on the cluster
reflects the cache and NVS size of that cluster. z/OS users will find that only the SETCACHE
CFW ON | OFF command is supported while other SETCACHE command options (example
DEVICE, SUBSYSTEM, DFW, NVS) are not accepted.
Migration considerations
The DS8000 is supported as an IBM 2105 for z/OS systems without the DFSMS and z/OS
small program enhancements (SPEs) installed. This allows customers to roll the SPE to each
system in a sysplex without having to take a sysplex-wide outage. An IPL will have to be taken
to activate the DFSMS and z/OS portions of this support.
Coexistance considerations
Support for the DS8000 running in 2105 mode on systems without this SPE installed is
provided. It consists of the recognition of the DS8000 real control unit type and device codes
when it runs in 2105 emulation on these down-level systems.
Input/Output definition files (IODF) created by HCD may be shared on systems that do not
have this SPE installed. Additionally, existing IODF files that define IBM 2105 control unit
records for a 2107 subsystem should be able to be used as long as 16 or fewer logical
subsystems are configured in the DS8000.
Each Parallel Access Volume (PAV) alias device must have an associated alias address
defined in the HCD/IOCP. Issue the ‘D M=CHP(xx)’ system command to verify that the aliases
are bound where they are expected to be. PAV aliases can be dynamically moved among
base addresses within the same LSS.
Wherever possible use dynamic PAV with Workload Manager to maximize the benefit of your
available aliases. The correct number of aliases for your workload can be determined from
analysis of RMF data. The PAV Tool available at
http://www-03.ibm.com/servers/eserver/zseries/zos/unix/bpxa1ty2.html#pavanalysis
In the absence of workload data to model consider the following rule of thumb: define as
many aliases as the number of FICON channels to the LSS multiplied by six times. Also, you
may use Table 16-1 on page 339 as a conservative recommendation for base to alias ratios.
1 - 3339 0.33 1
6679 - 10,017 1 3
23,374 - 30,051 2 6
50,086 - 60,102 3 9
More information regarding dynamic PAVs can be found on the Internet at:
http://www.ibm.com/s390/wlm/
RMF considerations
RMF reports all I/O activity against the Base PAV address —not by the Base and associated
Aliases. The performance information for the Base includes all Base and Alias activity.
Missing Interrupt Handler times for PAV alias addresses must not be set. An alias device
inherits the MIH of the base address to which it is bound and it is not possible to assign an
MIH value to an alias address. Alias devices are not known externally and are only known and
accessible by IOS. If an external method is used to attempt to set the MIH on an alias device
address, an IOS090I message will be generated. For example, the following message will be
observed for each attempt to set the MIH on an alias device:
IOS090I alias-device-number IS AN INVALID DEVICE
Tip: When setting MIH times in the IECIOSxx member of SYS1.PARMLIB, do not use
device ranges that include alias device numbers.
HyperPAV is supported starting with z/OS 1.8. For prior levels of operating system, the
support is provided as a small program enhancement (SPE) back to z/OS 1.6.
16.4.1 Connectivity
z/VM provides the following connectivity:
z/VM supports FICON attachment as 3990 Model 3 or 6 controller
Native controller mode 2105, 2107, and 1750 are supported on z/VM 5.2.0 with APAR
VM63952
– Brings support on par with z/OS
– z/VM simulates controller mode support by each guest
z/VM supports FCP attachment for Linux systems running as a guest.
z/VM itself supports FCP attached SCSI disks —starting with z/VM 5.1.0.
z/VM also provides the following support when using FCP attachment:
FCP attached SCSI LUNs as emulated 9336 Model 20 DASD
1 TB SCSI LUNs
Additional z/VM technical information for PAV support can be found on the z/VM Technical
Web site at: http://www.vm.ibm.com/storman/pav/.
Also, for further discussion on PAV see Section 10.4.9, “PAV in z/VM environments” on
page 191.
HyperPAV support
It is intention of IBM to provide HyperPAV support in the next z/VM release—note that all
statements regarding IBM plans, directions, and intent are subject to change or withdrawal
without notice.
16.4.4 MIH
z/VM sets its MIH value as a factor of 1.25 of what the hardware reports. This way, with the
DS8000 setting the MIH value to 30 seconds, z/VM will be set to a MIH value of approximately
37.5 seconds. This allows the guest to receive the MIH 30 seconds before z/VM does.
17
For further information on these topics, refer to the redbook iSeries and IBM TotalStorage: A
Guide to Implementing External Disk on eserver i5, SG24-7120.
17.1.1 Hardware
The DS8000 is supported on all System i models which support Fibre Channel attachment for
external storage. Fibre Channel was supported on all model 8xx onwards. AS/400 models
7xx and prior only supported SCSI attachment for external storage, so they cannot support
the DS8000.
There are three Fibre Channel adapters for System i. All support the DS8000:
2766 2 Gigabit Fibre Channel Disk Controller PCI
2787 2 Gigabit Fibre Channel Disk Controller PCI-X
5760 4 Gigabit Fibre Channel Disk Controller PCI-X
The System i Storage Web page provides information about current hardware requirements,
including support for switches. This can be found at:
http://www.ibm.com/servers/eserver/iseries/storage/storage_hw.html
17.1.2 Software
The System i must be running V5R2 or V5R3 (i5/OS) or later level of OS/400. In addition, at
the time of writing, the following PTFs were required:
V5R2: MF33327, MF33301, MF33469, MF33302, SI14711 and SI14754
V5R3: MF33328, MF33845, MF33437, MF33303, SI14690, SI14755 and SI14550
Prior to attaching the DS8000 to System i, you should check for the latest PTFs, which may
have superseded those shown here.
Table 17-1 on page 345 gives the number of extents required for different System i volume
sizes.
GiB represents “Binary Gigabytes” (230 bytes) and GB represents “Decimal Gigabytes” (109 bytes).
Note: Logical volumes of size 8.59 and 282.2 are not supported as System i Load Source
Unit (boot disk) where the Load Source Unit is to be located in the external storage server.
When creating the logical volumes for use with OS/400, you will see that in almost every
case, the OS/400 device size doesn’t match a whole number of extents, and so some space
will be wasted. You should also note that the #2766 and #2787 Fibre Channel Disk Adapters
used by System i can only address 32 LUNs, so creating more, smaller LUNs will require
more Input Output Adapters (IOAs) and their associated Input Output Processors (IOPs). For
more sizing guidelines for OS/400, refer to .
Under some circumstances, you may wish to mirror the OS/400 Load Source Unit (LSU) to a
LUN in the DS8000. In this case, only one LUN should be defined as unprotected; otherwise,
when mirroring is started to mirror the LSU to the DS8000 LUN, OS/400 will attempt to mirror
all unprotected volumes.
This will return the extents used for that volume into the Extent Pool. You will then be able to
create a new logical volume with the correct protection after a short period of time (depending
on the number of extents being returned to the extent pool). This is unlike ESS E20, F20 and
800 where the entire array containing the logical volume had to be reformatted.
However, before deleting the logical volume on the DS8000, you must first remove it from the
OS/400 configuration (assuming it was still configured). This is an OS/400 task which is
disruptive if the disk is in the System ASP or User ASPs 2-32 because it requires an IPL of
OS/400 to completely remove the volume from the OS/400 configuration. This is no different
from removing an internal disk from an OS/400 configuration. Indeed, deleting a logical
volume on the DS8000 is similar to physically removing a disk drive from an System i. Disks
can be removed from an Independent ASP with the IASP varied off without IPLing the
system.
Selection
3
3. Select Option 2, Work with disk configuration as shown in Figure 17-2 on page 347.
Selection
2
F3=Exit F12=Cancel
4. When adding disk units to a configuration, you can add them as empty units by selecting
Option 2 or you can choose to allow OS/400 to balance the data across all the disk units.
Normally, we recommend balancing the data. Select Option 8, Add units to ASPs and
balance data as shown in Figure 17-3.
Selection
8
F3=Exit F12=Cancel
5. Figure 17-4 on page 348 shows the Specify ASPs to Add Units to panel. Specify the ASP
number next to the desired units. Here we have specified ASP1, the System ASP. Press
Enter.
6. The Confirm Add Units panel will appear for review as shown in Figure 17-5. If everything
is correct, press Enter to continue.
Add will take several minutes for each unit. The system will
have the displayed protection after the unit(s) are added.
Serial Resource
ASP Unit Number Type Model Name Protection
1 Unprotected
1 02-89058 6717 074 DD004 Device Parity
2 68-0CA4E32 6717 074 DD003 Device Parity
3 68-0C9F8CA 6717 074 DD002 Device Parity
4 68-0CA5D96 6717 074 DD001 Device Parity
5 75-1118707 2107 A85 DD006 Unprotected
7. Depending on the number of units you are adding, this step could take some time. When it
completes, display your disk configuration to verify the capacity and data protection.
2. Expand the System i to which you wish to add the logical volume and sign on to that
server as shown in Figure 17-7.
3. Expand Configuration and Service, Hardware, and Disk Units as shown in Figure 17-8
on page 350.
4. You will be asked to sign on to SST as shown in Figure 17-9. Enter your Service tools ID
and password and press OK.
8. A confirmation panel like that shown in Figure 17-11 will appear to summarize the disk
pool configuration. Select Next to continue.
9. Now you need to add disks to the new disk pool. On the Add to disk pool screen, click the
Add disks button as shown in Figure 17-12 on page 352.
10.A list of non-configured units similar to that shown in Figure 17-13 appears. Highlight the
disks you want to add to the disk pool and click Add.
11.A confirmation screen appears as shown in Figure 17-14 on page 353. Click Next to
continue.
12.A summary of the Disk Pool configuration similar to Figure 17-15appears. Click Finish to
add the disks to the Disk Pool.
13.Take note of and respond to any message dialogs which appear. After taking action on any
messages, the New Disk Pool Status panel shown in Figure 17-16 on page 354 displays
and shows progress. This step may take some time, depending on the number and size of
the logical units being added.
15.The new Disk Pool can be seen on System i Navigator Disk Pools in Figure 17-18.
16.To see the logical volume, as shown in Figure 17-19 on page 355, expand Configuration
and Service, Hardware, Disk Pools and click the disk pool you just created.
17.5 Multipath
Multipath support was added for external disks in V5R3 of i5/OS (also known as OS/400
V5R3). Unlike other platforms which have a specific software component, such as Subsystem
Device Driver (SDD), multipath is part of the base operating system. At V5R3 and V5R4, up
to eight connections can be defined from multiple I/O adapters on an System i server to a
single logical volume in the DS8000. Each connection for a multipath disk unit functions
independently. Several connections provide availability by allowing disk storage to be utilized
even if a single path fails.
Multipath is important for System i because it provides greater resilience to SAN failures,
which can be critical to OS/400 due to the single level storage architecture. Multipath is not
available for System i internal disk units but the likelihood of path failure is much less with
internal drives. This is because there are fewer interference points where problems can occur,
such as long fiber cables and SAN switches, as well as the increased possibility of human
error when configuring switches and external storage, and the concurrent maintenance on the
DS8000 which may make some paths temporarily unavailable.
Many System i customers still have their entire environment in the System ASP and loss of
access to any disk will cause the system to fail. Even with User ASPs, loss of a UASP disk will
eventually cause the system to stop. Independent ASPs provide isolation such that loss of
disks in the IASP will only affect users accessing that IASP while the rest of the system is
unaffected. However, with multipath, even loss of a path to disk in an IASP will not cause an
outage.
Prior to multipath being available, some clients used OS/400 mirroring to two sets of disks,
either in the same or different external disk subsystems. This provided implicit dual-path as
long as the mirrored copy was connected to a different IOP/IOA, BUS, or I/O tower. However,
this also required two copies of data. Since disk level protection is already provided by
RAID-5 or RAID-10 in the external disk subsystem, this was sometimes seen as
unnecessary.
With the combination of multipath and RAID-5 or RAID-10 protection in the DS8000, we can
provide full protection of the data paths and the data itself without the requirement for
additional disks.
1. IO Frame
2. BUS
3. IOP
4. IOA 6. Port
7. Switch
5. Cable 8. Port
9. ISL
10. Port
11. Switch
12. Port
13. Cable
14. Host Adapter
15. IO Drawer
Unlike other systems, which may only support two paths (dual-path), OS/400 V5R3 supports
up to eight paths to the same logical volumes. As a minimum, you should use two, although
some small performance benefits may be experienced with more. However, since OS/400
multipath spreads I/O across all available paths in a round-robin manner, there is no load
balancing, only load sharing.
All can be used for multipath and there is no requirement for all paths to use the same type of
adapter. Both adapters can address up to 32 logical volumes. This does not change with
multipath support. When deciding how many I/O adapters to use, your first priority should be
to consider performance throughput of the IOA since this limit may be reached before the
maximum number of logical units. See “Sizing guidelines” on page 363 for more information
on sizing and performance guidelines.
Figure 17-22 on page 358 shows an example where 48 logical volumes are configured in the
DS8000. The first 24 of these are assigned via a host adapter in the top left I/O drawer in the
DS8000 to a Fibre Channel I/O adapter in the first System i I/O tower or rack. The next 24
logical volumes are assigned via a host adapter in the lower left I/O drawer in the DS8000 to
a Fibre Channel I/O adapter on a different BUS in the first System i I/O tower or rack. This
would be a valid single path configuration.
Volumes 1-24
Volumes 25-48
IO Drawers and
IO Drawer IO Drawer
IO Drawer IO Drawer
Host Adapters
Host Adapter 3 Host Adapter 4
BUS a BUS x
FC IOA FC IOA
iSeries IO
BUS b
FC IOA FC IOA
BUS y Towers/Racks
Logical connection
To implement multipath, the first group of 24 logical volumes is also assigned to a Fibre
Channel I/O adapter in the second System i I/O tower or rack via a host adapter in the lower
right I/O drawer in the DS8000. The second group of 24 logical volumes is also assigned to a
Fibre Channel I/O adapter on a different BUS in the second System i I/O tower or rack via a
host adapter in the upper right I/O drawer.
You will then be presented with a panel similar to Figure 17-23 on page 359. The values in the
Resource Name column show DDxxx for single path volumes and DMPxxx for those which
have more than one path. In this example, the 2107-A85 logical volume with serial number
75-1118707 is available through more than one path and reports in as DMP135.
4. Specify the ASP to which you wish to add the multipath volumes.
Note: For multipath volumes, only one path is shown. In order to see the additional
paths, see “Managing multipath volumes using System i Navigator” on page 360.
5. You are presented with a confirmation screen as shown in Figure 17-24. Check the
configuration details and if correct, press Enter to accept.
Add will take several minutes for each unit. The system will
have the displayed protection after the unit(s) are added.
Serial Resource
ASP Unit Number Type Model Name Protection
1 Unprotected
1 02-89058 6717 074 DD004 Device Parity
2 68-0CA4E32 6717 074 DD003 Device Parity
3 68-0C9F8CA 6717 074 DD002 Device Parity
4 68-0CA5D96 6717 074 DD001 Device Parity
5 75-1118707 2107 A85 DMP135 Unprotected
Follow the steps outlined in “Adding volumes to an Independent Auxiliary Storage Pool” on
page 348.
When you get to the point where you will select the volumes to be added, you will see a panel
similar to that shown in Figure 17-25. Multipath volumes appear as DMPxxx. Highlight the
disks you want to add to the disk pool and click Add.
Note: For multipath volumes, only one path is shown. In order to see the additional paths,
see “Managing multipath volumes using System i Navigator” on page 360.
The remaining steps are identical to those in “Adding volumes to an Independent Auxiliary
Storage Pool” on page 348.
When using the standard disk panels in System i Navigator, only a single (the initial) path is
shown. The following steps show how to see the additional paths.
To see the number of paths available for a logical unit, open System i Navigator and expand
Configuration and Service, Hardware, and Disk Units as shown in Figure 17-26 on
page 361 and click All Disk Units. The number of paths for each unit is shown in column
Number of Connections visible on the right of the panel. In this example, there are 8
connections for each of the multipath units.
To see the other connections to a logical unit, right click on the unit and select Properties, as
shown in Figure 17-26.
You now get the General Properties tab for the selected unit, as shown in Figure 17-27. The
first path is shown as Device 1 in the box labelled Storage.
To see the other paths to this unit, click the Connections tab, as shown in Figure 17-28,
where you can see the other seven connections for this logical unit.
When you use multipath disk units, you must consider the implications of moving IOPs and
multipath connections between nodes. You must not split multipath connections between
nodes, either by moving IOPs between logical partitions or by switching expansion units
between systems. If two different nodes both have connections to the same LUN in the
DS8000, both nodes might potentially overwrite data from the other node.
The system enforces the following rules when you use multipath disk units in a
multiple-system environment:
If you move an IOP with a multipath connection to a different logical partition, you must
also move all other IOPs with connections to the same disk unit to the same logical
partition.
When you make an expansion unit switchable, make sure that all multipath connections to
a disk unit will switch with the expansion unit.
When you configure a switchable independent disk pool, make sure that all of the required
IOPs for multipath disk units will switch with the independent disk pool.
If a multipath configuration rule is violated, the system issues warnings or errors to alert you
of the condition. It is important to pay attention when disk unit connections are reported
missing. You want to prevent a situation where a node might overwrite data on a LUN that
belongs to another node.
Disk unit connections might be missing for a variety of reasons, but especially if one of the
preceding rules has been violated. If a connection for a multipath disk unit in any disk pool is
found to be missing during an IPL or vary on, a message is sent to the QSYSOPR message
queue.
If a connection is missing, and you confirm that the connection has been removed, you can
update Hardware Service Manager (HSM) to remove that resource. Hardware Service
Manager is a tool for displaying and working with system hardware from both a logical and a
packaging viewpoint, an aid for debugging Input/Output (I/O) processors and devices, and for
fixing failing and missing hardware. You can access Hardware Service Manager in System
Service Tools (SST) and Dedicated Service Tools (DST) by selecting the option to start a
service tool.
Performance Workload
Tools Reports description
Workload Workload
statistics characteristics
Other
requirements:
HA, DR etc.
Rules of thumb
SAN Fabric
Proposed
configuration
Workload from
other servers
Modeling with
Adjust config
Disk Magic
based on DM
modeling
Requirements and
expectations met ?
No
Yes
Finish
For workloads with critical response times, you may not want to use all the capacity in an
array. For 73 GB DDMs you may plan to use about 300 GB capacity per 8 drive array. The
remaining capacity could possibly be used for infrequently accessed data. For example, you
may have archive data, or some data such as images, which is not accessed regularly, or
perhaps FlashCopy target volumes which could use this capacity, but not impact on the
I/O /sec on those arrays.
For very high write environments, you may also consider using RAID-10, which offers a higher
I/O rate per GB than RAID-5 as shown in Table 17-3 on page 366. However, the majority of
System i workloads do not require this.
17.6.2 Cache
In general, System i workloads do not benefit from large cache. Still, depending on the
workload (as shown in OS/400 Performance Tools System, Component and Resource
Interval reports) you may see some benefit in larger cache sizes. However, in general, with
large System i main memory sizes, OS/400 Expert Cache can reduce the benefit of external
cache.
Since this guideline is based only on System i adapters and Access Density (AD) of System i
workload, it doesn't change when using the DS8000.
Note: Access Density is the ratio that results from dividing the occupied disk space capacity
by the average I/Os per second. These values can be obtained from the OS/400 System,
Component and Resource Interval performance reports.
Table 17-2 shows the approximate capacity which can be supported with various IOA/IOP
combinations.
Note: ***Size the same capacity per adapter 5760 as for 2787 on 2844. For transfer sizes
larger than 16KB, size about 50% more capacity than for adapter 2787.
For most System i workloads, Access Density is usually below 2, so if you do not know it, the
Rule of thumb column is a typical value to use.
From the values in Table 17-2, you can calculate the number of System i Fibre Channel
adapters for your required System i disk capacity. As each I/O adapter can support a
maximum of 32 LUNs, divide the capacity per adapter by 32 to give the approximate average
size of each LUN.
For example, assume you require 2 TB capacity and are using 2787 I/O adapters with 2844
I/O processors. If you know the access density, calculate the capacity per I/O adapter, or use
the rule-of-thumb. Let’s assume the rule-of-thumb of 500 GB per adapter. In this case, we
would require four I/O adapters to support the workload. If we were able to have variable
LUNs sizes, we could support 32 15.6 GB LUNs per I/O adapter. However, since OS/400 only
supports fixed volume sizes, we could support 28 17. 5 GB volumes to give us approximately
492 GB per adapter.
When considering the number of ranks, take into account the maximum disk operations per
second per rank as shown in Table 17-3. These are measured at 100% DDM Utilization with
no cache benefit and with the average I/O being 4KB. Larger transfer sizes will reduce the
number of operations per second.
Based on these values you can calculate how many host I/O per second each rank can
handle at the recommended utilization of 40%. This is shown for workload read-write ratios of
70% read and 50% read in Table 17-3.
As can be seen in Table 17-3, RAID-10 can support higher host I/O rates than RAID-5.
However, you must balance this against the reduced effective capacity of a RAID-10 rank
when compared to RAID-5.
However, you may consider sharing ranks when the other servers’ workloads have a
sustained low disk I/O rate compared to the System i I/O rate. Generally, System i has a
relatively high I/O rate while that of other servers may be lower – often below one I/O per GB
per second.
As an example, a Windows file server with a large data capacity may normally have a low I/O
rate with less peaks and could be shared with System i ranks. However, SQL, DB or other
application servers may show higher rates with peaks, and we recommend using separate
ranks for these servers.
Unlike its predecessor the ESS, capacity used for logical units on the DS8000 can be reused
without reformatting the entire array. Now, the decision to mix platforms on an array is only
one of performance, since the disruption previously experienced on ESS to reformat the array
no longer exists.
Based on available measurements and experiences with the ESS 800 we recommend you
should plan no more than four System i I/O adapters to one host port in the DS8000.
For a current list of switches supported under OS/400, refer to the System i Storage Web site
at: http://www-1.ibm.com/servers/eserver/iseries/storage/storage_hw.html.
17.7 Migration
For many System i customers, migrating to the DS8000 will be best achieved using traditional
Save/Restore techniques. However, there are some alternatives you may wish to consider.
Once an external logical volume has been created, it will always keep its model type and be
either protected or unprotected. Therefore, once a logical volume has been defined as
unprotected to allow it to be the mirror target, it cannot be converted back to a protected
model, and therefore will be a candidate for all future OS/400 mirroring, whether you want this
or not.
Consider the example shown in Figure 17-30 on page 368. Here, the System i has its internal
Load Source Unit (LSU) and possibly some other internal drives. The ESS provides additional
storage capacity. Using Metro Mirror or Global Copy, it is possible to create copies of the ESS
logical volumes in the DS8000.
When ready to migrate from the ESS to the DS8000, you should do a complete shutdown of
the System i, unassign the ESS LUNs and assign the DS8000 LUNs to the System i. After
IPLing the System i, the new DS8000 LUNs will be recognized by OS/400, even though they
are different models and have different serial numbers.
Note: It is important to ensure that both the Metro Mirror or Global Copy source and target
copies are not assigned to the System i at the same time because this is an invalid
configuration. Careful planning and implementation is required to ensure this does not
happen, otherwise unpredictable results may occur.
Me
tro
Mir
ror
The same setup can also be used if the ESS LUNs are in an IASP, although the System i
would not require a complete shutdown since varying off the IASP in the ESS, unassigning
the ESS LUNs, assigning the DS8000 LUNs and varying on the IASP would have the same
effect.
Clearly, you must also take into account the licensing implications for Metro Mirror and Global
Copy.
Note: This is a special case of using Metro Mirror or Global Copy and will only work if the
same System i is used, along with the LSU to attach to both the original ESS and the new
DS8000. It is not possible to use this technique to a different System i.
You can then use the OS/400 command STRASPBAL TYPE(*ENDALC) to mark the units to
be removed from the configuration as shown in Figure 17-31 on page 369. This can reduce
the down time associated with removing a disk unit. This will keep new allocations away from
the marked units.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
When you subsequently run the OS/400 command STRASPBAL TYPE(*MOVDTA) all data
will be moved from the marked units to other units in the same ASP, as shown in
Figure 17-32. Clearly you must have sufficient new capacity to allow the data to be migrated.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
You can specify a time limit that the function is to run for each ASP being balanced or the
balance can be set to run to completion. If the balance function needs to be ended prior to
this, use the End ASP Balance (ENDASPBAL) command. A message will be sent to the
system history (QHST) log when the balancing function is started for each ASP. A message
will also be sent to the QHST log when the balancing function completes or is ended.
If the balance function is run for a few hours and then stopped, it will continue from where it
left off when the balance function restarts. This allows the balancing to be run during off hours
over several days.
In order to finally remove the old units from the configuration, you will need to use Dedicated
Service Tools (DST) and re-IPL the system (or partition).
Using this method allows you to remove the existing storage units over a period of time.
However, it does require that both the old and new units are attached to the system at the
same time so it may require additional IOPs and IOAs if migrating from an ESS to a DS8000.
It may be possible in your environment to re-allocate logical volumes to other IOAs, but
careful planning and implementation will be required.
Boot from SAN support enables you to take advantage of some of the advanced features
available with the DS8000 series and Copy Services functions. One of these functions is
known as FlashCopy; this function allows you to perform a near instantaneous copy of the
data held on a LUN or group of LUNs. Therefore, when you have a system that only has
external LUNs with no internal drives, you are able to create a clone of your system.
Important: When we refer to a clone, we are referring to a copy of a system that only uses
external LUNs. Boot (or IPL) from SAN is therefore a prerequisite for this.
The following i5 I/O adapters are required to attach a DS8000 directly to an i5 AIX partition:
0611 Direct Attach 2 Gigabit Fibre Channel PCI
0625 Direct Attach 2 Gigabit Fibre Channel PCI-X
It is also possible for the AIX partition to have its storage virtualized, whereby a partition
running OS/400 hosts the AIX partition's storage requirements. In this case, if using DS8000,
they would be attached to the OS/400 partition using either of the following I/O adapters:
2766 2 Gigabit Fibre Channel Disk Controller PCI
For more information on running AIX in an i5 partition, refer to the i5 Information Center at:
http://publib.boulder.ibm.com/infocenter/iseries/v1r2s/en_US/index.htm?info/iphat
/iphatlparkickoff.htm
Note: AIX will not run in a partition on earlier 8xx and prior System i hosts.
The DS8000 requires the following System i I/O adapters to attach directly to an System i or
i5 Linux partition.
0612 Linux Direct Attach PCI
0626 Linux Direct Attach PCI-X
It is also possible for the Linux partition to have its storage virtualized, where a partition
running i5/OS hosts the Linux partition's storage. This storage can be made up of any
supported storage, such as a mix of internal storage and DS8000s. To use the DS8000 for
this hosted storage running under the i5/OS partition, use either of the following I/O adapters:
2766 2 Gigabit Fibre Channel Disk Controller PCI
2787 2 Gigabit Fibre Channel Disk Controller PCI-X
More information on running Linux in an System i partition can be found in the System i
Information Center at:
V5R2 at: http://publib.boulder.ibm.com/iseries/v5r2/ic2924/index.htm.
V5R3 at: http://publib.boulder.ibm.com/infocenter/iseries/v5r3/ic2924/index.htm.
Part 5
18
When IBM releases new microcode for the DS8000, it is released in the form of a bundle. The
term bundle is used because a new code release may include updates for various DS8000
components. These updates are tested together and then the various code packages are
bundled together into one unified release. In general, when referring to what code level is
being used on a DS8000, the term bundle should therefore be used. However, components
within the bundle will have their own revision levels. For instance, bundle 6.0.500.53 contains
DS CLI Version 5.0.500.134 and DS Storage Manager Version 5.0.5.0302. It is important that
you always match your DS CLI version to the bundle installed on your DS8000.
While the installation process described above may seem complex, it will not require a great
deal of user intervention. The code installer will normally just start the process and then
monitor its progress using the S-HMC.
Attention: An upgrade of your DS8000 microcode may require that you upgrade the level
of your DS CLI version and may also require that you upgrade your S-HMC version. Check
with your IBM representative on the description and contents of the release bundle.
There is also the ability to install microcode update bundles non-concurrently, with all
attached hosts shut down. However, this should not be necessary. This method is usually only
employed at DS8000 installation time.
The HMC portion of the bundle should be updated one to two days before the rest of the
bundle is installed. This is because prior to the update of the SFI, a pre-verification process is
run to ensure that no pre-existing conditions exist that need to be corrected before the SFI
code update. The HMC code update will update the pre-verification routine. Thus, we should
update the HMC code and then run the updated pre-verification routine. If problems are
detected, we still have one to two days before the scheduled code installation window to
correct them.
brought online and the process moves to the next ESCON adapter. The use of CUIR removes
any need for operator intervention. CUIR is not used for remote mirror and copy paths.
CUIR allows the DS8000 to request that all attached system images set all paths required for
a particular service action to the offline state. System images with the appropriate level of
software support will respond to such requests by varying off the affected paths, and either
notifying the DS8000 subsystem that the paths are offline, or that it cannot take the paths
offline. CUIR reduces manual operator intervention and the possibility of human error during
maintenance actions, at the same time reducing the time required for the maintenance. This
is particularly useful in environments where there are many systems attached to a DS8000.
When it comes time to copy the new code bundle onto the DS8000, there are two ways to
achieve this:
Load the new code bundle onto the HMC using CDs.
Download the new code bundle directly from IBM using FTP.
The ability to download the code from IBM eliminates the need to order or burn CDs.
However, it may require firewall changes to allow the HMC to connect using FTP to the site
listed above.
Note that the actual time required for the concurrent code load will vary based on the bundle
you are currently running and the bundle you are going to. It is not possible to state here how
long updates will take. Always consult with your IBM Service Representative.
The DS8000 series includes many enhancements to make the Licensed Machine Code
change process simpler, quicker, and more automated. If you would prefer, you may request
IBM to install downloadable Licensed Machine Code changes; however, you may be charged
for that service.
19
SNMP provides a standard MIB that includes information such as IP addresses and the
number of active TCP connections. The actual MIB definitions are encoded into the agents
running on a system.
MIB-2 is the Internet standard MIB that defines over 100 TCP/IP specific objects, including
configuration and statistical information such as:
Information about interfaces
Address translation
IP, ICMP (Internet-control message protocol), TCP, and UDP
SNMP can be extended through the use of the SNMP Multiplexing protocol (the SMUX
protocol) to include enterprise-specific MIBs that contain information related to a specific
environment or application. A management agent (a SMUX peer daemon) retrieves and
maintains information about the objects defined in its MIB, and passes this information on to a
specialized network monitor or network management station (NMS).
The SNMP protocol defines two terms, agent and manager, instead of the client and server
used in many other TCP/IP protocols:
Agents send traps to the SNMP manager to indicate that a particular condition exists on the
agent system, such as the occurrence of an error. In addition, the SNMP manager generates
traps when it detects status changes or other unusual conditions while polling network
objects.
SNMP defines six generic types of traps and allows definition of enterprise-specific traps. The
trap structure conveys the following information to the SNMP manager:
Agent’s object that was affected
IP address of the agent that sent the trap
Event description (either a generic trap or enterprise-specific trap, including trap number)
Time stamp
Optional enterprise-specific trap identification
List of variables describing the trap
You can gather various information on the specific IP hosts by sending the SNMP get and
get-next request, and can update the configuration of IP hosts by sending the SNMP set
request.
The SNMP agent can send SNMP trap requests to SNMP managers, which listen on UDP
port 162. The SNMP trap1 requests sent from SNMP agents can be used to send warning,
alert, or error notification messages2 to SNMP managers.
Note that you can configure an SNMP agent to send SNMP trap requests to multiple SNMP
managers. Figure 19-1 illustrates the characteristics of SNMP architecture and
communication.
Therefore, you should be careful about the SNMP security. At the very least:
Do not use the default community name (public and private).
Do not allow access to hosts that are running the SNMP agent, from networks or IP hosts
that do not necessarily require access.
You may want to physically secure the network to which you would send SNMP packets by
using a firewall, because community strings are included as plain text in SNMP packets.
Most hardware and software vendors provide you with extended MIB objects to support their
own requirements. The SNMP standards allow this extension by using the private sub-tree,
called enterprise specific MIB. Because each vendor has unique MIB sub-tree under the
private sub-tree, there is no conflict among vendor original MIB extension.
A trap message contains pairs of an OID and a value shown in Table 19-1 to notify the cause
of the trap message. You can also use type 6, the enterpriseSpecific trap type, when you
have to send messages that are not fit for other predefined trap types, for example, DISK I/O
error and application down. You can also set an integer value field called Specific Trap on
your trap message.
The DS8000 does not have an SNMP agent installed that can respond to SNMP polling. The
SNMP Community Name can be set within the DS Storage Manager GUI or the DS CLI. The
default Community Name is set to public.
The management server that is configured to receive the SNMP traps will receive all the
generic trap 6 and specific trap 3 messages, which will be sent out in parallel with the Call
Home to IBM.
Before configuring SNMP for the DS8000, you are required to get the destination address for
the SNMP trap and also the port information on which the Trap Daemon will listen.
A serviceable event will be posted as a generic trap 6 specific trap 3 message. The specific
trap 3 is the only event that is being sent out for serviceable events. For reporting Copy
Services events generic trap 6 and specific traps 100, 101, 102, 200, 202, 210, 211, 212, 213,
214, 215, 216, or 217 will be sent out.
The SNMP trap will be sent out in parallel to a Call Home for service to IBM.
For open events in the event log, a trap will be sent out every 8 hours until the event is closed.
This chapter describes only the messages and the circumstances when sent out by the
DS8000. For detailed information about these functions and terms, refer to IBM System
Storage DS8000 Series: Copy Services with System z servers, SG24-6787 and IBM System
Storage DS8000 Series: Copy Services in Open Environments, SG24-6788.
If one or several links (but not all links) are interrupted, a trap 100, as shown in Example 19-2,
is posted and will indicate that the redundancy is degraded. The RC column in the trap will
represent the reason code for the interruption of the link — reason codes are listed in
Table 19-2 on page 387.
If all links all interrupted, a trap 101, as shown in Example 19-3, is posted. This event
indicates that no communication between the primary and the secondary system is possible
any more.
Once the DS8000 can communicate again via any of the links, trap 102, as shown in
Example 19-4, is sent once one or more of the interrupted links are available again.
00 No path.
02 Initialization failed. ESCON link reject threshold exceeded when attempting to send ELP or RID
frames.
07 Secondary CU SS ID mismatch or failure of the I/O that collects secondary information for validation.
08 ESCON link is offline. This is caused by the lack of light detection coming from a host, peer, or switch.
0A The primary control unit port or link cannot be converted to channel mode since a logical path is
already established on the port or link. The establish paths operation will not be retried within the
control unit automatically.
11 Reserved.
12 Reserved.
16 Fibre Channel Path Secondary Adapter not PPRC capable. This could be due to:
Secondary Adapter not configured properly, or does not have the correct microcode loaded.
The secondary adapter is already a target of 32 different ESSs, DS8000s, or DS6000s.
Group option, will be sent. This trap could be handled by automation software like eRCMF to
freeze this Consistency Group.
Example 19-5 Trap 200: LSS-Pair Consistency Group remote mirror pair error
LSS-Pair Consistency Group PPRC-Pair Error
UNIT: Mnf Type-Mod SerialNm LS LD SR
PRI: IBM 2107-922 75-03461 56 84 08
SEC: IBM 2107-9A2 75-ABTV1 54 84
Trap 202, as shown in Example 19-6, will be sent out if a remote Copy Pair goes into a
suspend State. The trap contains the serial number (SerialNm) of the primary and secondary
machine, the LSS (LS), and the logical device (LD). To avoid SNMP trap flooding, the number
of SNMP traps for the LSS will be throttled. The complete suspended pair information is
represented in the summary. The last row of the trap represents the suspend state for all pairs
in the reporting LSS. The suspended pair information contains a hexadecimal string of a
length of 64 characters. By converting this hex string into binary, each bit will represent a
single device. If the bit is 1 then the device is suspended; otherwise the device is still in full
duplex mode.
Example 19-6 Trap 202: Primary remote mirror devices on LSS suspended due to error
Primary PPRC Devices on LSS Suspended Due to Error
UNIT: Mnf Type-Mod SerialNm LS LD SR
PRI: IBM 2107-922 75-20781 11 00 03
SEC: IBM 2107-9A2 75-ABTV1 21 00
Start: 2005/11/14 09:48:05 CST
PRI Dev Flags (1 bit/Dev, 1=Suspended):
C000000000000000000000000000000000000000000000000000000000000000
Trap 210, as shown in Example 19-7, is sent out when a Consistency Group in a Global Mirror
environment was successfully formed.
Example 19-7 Trap210: Global Mirror initial Consistency Group successfully formed
2005/11/14 15:30:55 CET
Asynchronous PPRC Initial Consistency Group Successfully Formed
UNIT: Mnf Type-Mod SerialNm
IBM 2107-922 75-20781
Session ID: 4002
Trap 211, as shown in Example 19-8, will be sent out if the Global Mirror setup got into an
severe error state, where no attempts will be done to form a Consistency Group.
Trap 212, as shown in Example 19-9 on page 389, is sent out when a Consistency Group
could not be created in a Global Mirror relation. Some of the reasons could be:
Volumes have been taken out of a copy session.
The remote copy link bandwidth might not be sufficient.
The FC link between the primary and secondary system is not available.
Example 19-9 Trap 212: Global Mirror Consistency Group failure - Retry will be attempted
Asynchronous PPRC Consistency Group Failure - Retry will be attempted
UNIT: Mnf Type-Mod SerialNm
IBM 2107-922 75-20781
Session ID: 4002
Trap 213, as shown in Example 19-10, will be sent out when a Consistency Group in a Global
Mirror environment could be formed after a previous Consistency Group formation failure.
Example 19-10 Trap 213: Global Mirror Consistency Group successful recovery
Asynchronous PPRC Consistency Group Successful Recovery
UNIT: Mnf Type-Mod SerialNm
IBM 2107-9A2 75-ABTV1
Session ID: 4002
Trap 214, as shown in Example 19-11, will be sent out if a Global Mirror Session is terminated
using the DS CLI command rmgmir or the corresponding GUI function.
Trap 215, as shown in Example 19-12, will be sent out if in the Global Mirror Environment the
Master has detected a failure to complete the FlashCopy commit. The trap will be sent out
after a number of commit retries have failed.
Example 19-12 Trap 215: Global Mirror FlashCopy at remote site unsuccessful
Asynchronous PPRC FlashCopy at Remote Site Unsuccessful
A UNIT: Mnf Type-Mod SerialNm
IBM 2107-9A2 75-ABTV1
Session ID: 4002
Trap 216, as shown in Example 19-13, will be sent out if a Global Mirror Master cannot
terminate the Global Copy relationship at one of his Subordinates (slave). This might occur if
the Master is terminated with rmgmir but the Master cannot terminate the copy relationship on
the Subordinate. The user may need to run a rmgmir against the Subordinate to prevent any
interference with other Global Mirror sessions.
Trap 217, as shown in Example 19-14 on page 390, will be sent out if a Global Mirror
environment was suspended by the DS CLI command pausegmir or the corresponding GUI
function.
The network management server that is configured on the HMC will receive all the generic
trap 6 specific trap 3 messages, which will be sent out in parallel with any to the Call Home to
IBM.
dscli> showsp
Date/Time: November 16, 2005 10:15:04 AM CET IBM DSCLI Version: 5.1.0.204
Name IbmStoragePlex_2
desc ATS #1
acct -
SNMP Enabled
SNMPadd 10.10.10.11,10.10.10.12
emailnotify Disabled
emailaddr -
emailrelay Disabled
emailrelayaddr -
emailrelayhost -
20
The storage complex uses the Call Home method to send heartbeat information to IBM, and
by doing this ensures that the DS HMC is able to initiate a Call Home to IBM in the case of an
error. Should the heartbeat information not reach IBM, a service call to the client is initiated by
IBM to investigate the status of the DS8000. The Call Home can either be configured for
modem or Internet setup. The Call Home service can only be initiated by the DS HMC.
For the connection, a Virtual Private Internet (VPN) tunnel is established using the Internet
Security Architecture (IPSec) protocol. The VPN tunnel is established using an IPSec
end-to-end IPSec connection between the HMC and the IBM network.
Allowing the HMC to also set up a VPN connection to IBM adds additional redundancy to your
environment, as the HMC can use the VPN tunnel via the Ethernet to do the Call Home to
IBM or use the modem line, in case the Call Home over the network cannot be established.
Tip: For detailed information, IP addresses, and port information about remote support
using an Internet connection, refer to the document VPN Security and Implementation,
which is available in the DS8000 Document Library under the topic VPN implementation at:
http://www-03.ibm.com/servers/storage/support/disk/ds8100/installing.html
or directly access this URL to download the PDF document:
http://www-1.ibm.com/support/docview.wss?&rs=1114&uid=ssg1S1002693
20.2 Connections
The remote support can either be configured for modem or Internet setup.
Note: IBM recommends that the DS HMC be connected to the client’s public network over
a secure VPN connection, instead of a dial-up connection.
Public Telephone
Async Modem
Retain Network
FIREWALL
IBM VPN
FIREWALL
Server
Figure 20-2 shows the establish process for a VPN connection. When the IBM support
personnel needs to connect to the DS8000, support will call in using a modem connection
and an ASCII terminal screen will be presented. This ASCII screen will request a one-time
password, which is generated with a presented key and an IBM system that is calculating this
one-time password. Once the access is granted, the support user establishes a VPN tunnel
back to IBM.
HMC
Support/PFE
2. VPN tunnel established back from HMC
Public Telephone
Async Modem
RS3 Network
FIREWALL
IBM VPN
FIREWALL
Server
Public Telephone
Async Modem
RS3
Network
FIREWALL
IBM VPN
Server
IBM
FIREWALL
AT&T Global
Intranet AT&T OR
Network
Client
Support/PFE LAN
Internet HMC
Depending on the outbound configuration made during the setup of the HMC, the HMC uses
the modem or the network connection to establish the VPN tunnel. The support person then
needs to log into an IBM system, which acts as the other end of the VPN tunnel. Now only the
workstation from where the VPN tunnel was opened from can connect to this VPN tunnel.
The tunnel will automatically stop if the support person logs out or the VPN tunnel was not
opened by IBM support within 10 minutes.
Developer Allowed root access, using ssh, but only if a PE user is logged in via
WebSM.
Client
Network
DMZ
DS8000
Subsystems
Integrated
Firewall Phoneline
Proxy
DMZ
IBM
Network
Optional
Firewall
eth eth
Client
eth modem Network
DMZ
DS8000
VPN
Subsystems Integrated
Firewall
Proxy
Redundant DS HMC
Internet
Ethernet eth eth
Fabric
eth modem
Integrated Phoneline
Firewall DMZ
Proxy
IBM
Network
21
The disk drives are installed in storage enclosures. A storage enclosure interconnects the
DDMs to the Fibre Channel switch cards that connect to the device adapters. Each storage
enclosure contains a redundant pair of Fibre Channel switch cards. See Figure 21-1.
Storage Enclosure
DDM
The storage enclosures are always installed in pairs, one enclosure in the front of the unit and
one enclosure in the rear. A storage enclosure pair can be populated with one or two disk
drive sets (16 or 32 DDMs). All DDMs in a disk enclosure pair must be of the same type
(capacity and speed). If a disk enclosure pair is populated with only 16 DDMs, disk drive filler
modules will be installed in the vacant DDM slots. This is to maintain correct cooling airflow
throughout the enclosure.
Each storage enclosure attaches to two separate device adapters (DAs). The DAs are the
raid adapter cards that connect the processor complexes to the DDMs. The DS8000 DA
cards always come as a redundant pair, so we refer to them as DA pairs.
Physical installation and testing of the device adapters, storage enclosure pairs, and DDMs is
performed by your IBM Service Representative. After the additional capacity is added
successfully, the new storage will appear as additional un-configured array sites.
You may need to obtain new license keys and apply them to the storage image before you
start configuring the new capacity; see Chapter 11, “Features and license keys” on page 197.
You will not be able to create ranks using the new capacity if this causes your machine to
exceed its license key limits.
As a general rule, when adding capacity to a DS8000, storage hardware is populated in the
following order:
1. DDMs are added to under-populated enclosures. Whenever you add 16 DDMs to a
machine, eight DDMs are installed into the front storage enclosure and eight into the rear
storage enclosure.
2. Once the first storage enclosure pair on a DA pair is fully populated with DDMs (32 DDMs
total), a second storage enclosure pair can be added to that DA pair.
3. Once a DA pair has two fully populated storage enclosure pairs (64 DDMs total), we add
another DA pair. The base frame supports two DA pairs. The first expansion frame
supports an additional two DA pairs for model 931 (for a total of four DA pairs), or an
additional four DA pairs for a model 932 or 9B2 (for a total of six DA pairs). In addition,
models 932 and 9B2 can have a second expansion frame that supports an additional two
DA pairs (for a total of 8 pairs). The DA cards are installed into the I/O enclosures that are
located at the bottom of the racks. They are not located in the storage enclosures.
4. The last stage varies according to model:
– For a model 931, once we have installed four DA pairs, and each DA pair has two fully
populated storage enclosure pairs (that is, we have 4 DA pairs with 64 DDMs on each
pair), then we start adding extra storage enclosure pairs to two of the DA pairs (the DA
pairs numbered 2 and 0). We can do this till these DA pairs have four storage
enclosure pairs each (128 DDMs per DA pair). At this point two of the DA pairs have 64
DDMs and two of the DA pairs have 128 DDMs. Both racks are now fully populated and
no more capacity can be added.
– For a model 932 or 9B2, once we have installed eight DA pairs, and each DA pair has
two fully populated storage enclosure pairs (that is 8 DA pairs with 64 DDMs on each
pair), then we start adding extra storage enclosure pairs to two of the DA pairs (the DA
pairs numbered 2 and 0). We can do this till these DA pairs have four storage
enclosure pairs each (128 DDMs per DA pair). At this point six of the DA pairs have 64
DDMs and two of the DA pairs have 128 DDMs. All three racks are now fully populated
and no more capacity can be added.
In Example 21-1 on page 400 we use these commands to list the DAs, the SEs, the DDMs,
and the array sites in a DS8000. Because there are 64 DDMs in the example machine, only
part of the list of DDMs is shown. If you use the -l parameter for these commands you will
receive additional information. In this example we have one DA pair (two device adapters) and
two storage enclosure pairs (four storage enclosures). Because we have 64 DDMs, we have
eight array sites (since each array site consists of eight DDMs). In this example six of the
array sites are in use, and two are Unassigned (meaning that no array is using that array site).
There are various rules about CoD and these are explained in the IBM System Storage
DS8000 Introduction and Planning Guide, GC35-0515. The purpose of this chapter is to
explain aspects of implementing a DS8000 that has CoD disk packs.
In many database environments, it is not unusual to have very rapid growth in the amount of
disk space required for your business. This can create a problem if there is an unexpected
and urgent need for disk space and no time to raise a purchase order or wait for the disk to be
delivered.
With this offering, up to four Standby CoD disk drive sets (64 disk drives) can be factory or
field installed into your system. To activate, you simply logically configure the disk drives for
use —a non disruptive activity that does not require intervention from IBM. Upon activation of
any portion of a Standby CoD disk drive set, you must place an order with IBM to initiate
billing for the activated set. At that time, you can also order replacement CoD disk drive sets.
This offering allows you to purchase licensed functions based upon your machines physical
capacity excluding un-configured Standby CoD capacity. This can help improve your cost of
ownership since your extent of IBM authorization for licensed functions can grow at the same
time you need your disk capacity to grow.
Contact your IBM representative to obtain additional information regarding Standby CoD
offering terms and conditions.
Now using the signature, log on to the DSFA Web site at:
https://www-306.ibm.com/storage/dsfa/index.jsp.
On the View Machine Summary tab you will see whether the CoD indicator is on or off. In
Figure 21-2 you can see an example of a 2107-921 machine that has the CoD indicator.
If instead you see 0900 Non-Standby CoD then the CoD feature has not been ordered for
your machine.
Note: Older levels of DS8000 microcode may show the OEL key where CoD is present as
being 9223372 TB instead of 9999999 TB. This is not a problem. CoD is still present.
How a user can tell how many CoD array sites have been ordered
From the machine itself, there is no way to tell how many of the array sites in a machine are
CoD array sites versus array sites you can start using right away. During the machine order
process this should be clearly understood and documented.
These commands are available on every system supported for DS8000 attachment, but work
only with data organized on file systems. Data can be copied between file systems with
different sizes. Therefore, these methods can be used for consolidation of small volumes into
large ones.
The most significant disadvantage of this method is the disruptiveness. To preserve data
consistency, the applications writing to the data that is migrated have to be interrupted for the
duration of the copy process. Furthermore, some copy commands cannot preserve advanced
metadata such as access control lists or permissions.
Important: If your storage systems are attached through multiple paths, make sure that
the multipath drivers for the old storage and the new storage can coexist on the same
system. If not, you have to revert the host to a single path configuration and remove the
incompatible driver from the system before you attach the new storage system.
This method also requires the disruption of the applications writing to the data for the
complete process.
The LUNs provided by a DS8000 appear to the LVM or LDM as physical SCSI disks.
Usually the process is to set up a mirror of the data on the old disks to the new LUNs, wait
until it is synchronized, and split it at the cut over time. Some volume management software
provides commands that automate this process.
The biggest advantage of using the volume management software for data migration is that
the process can be totally non-disruptive, as long as the operating system allows you to add
and remove devices dynamically. Due to the virtualization nature of volume management
software, it also allows for all kinds of consolidation.
The major disadvantage of the volume management software mirroring method is that it
requires a lot of system administration intervention and attention. Production data is
manipulated while production is running, and it requires host cycles when the synchronization
of the data is running.
Attention: If you are planning to use volume management software functions to migrate
the data, be careful with some limitations such, as total number of physical disks in the
same volume group or volume set and, if you are consolidating volumes with different
sizes, check the procedures to see if this is possible.
The content of the volume group is shown with the lsvg -l command. Note that in the
column LPs and PPs the proportion is 1 by 1 that means that we have only one physical
copy of each logical data:
root:/ > lsvg -l rootvg
rootvg:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
hd5 boot 1 1 1 closed/syncd N/A
hd6 paging 8 8 1 open/syncd N/A
hd8 jfs2log 1 1 1 open/syncd N/A
hd4 jfs2 1 1 1 open/syncd /
hd2 jfs2 21 21 1 open/syncd /usr
hd9var jfs2 1 1 1 open/syncd /var
hd3 jfs2 1 1 1 open/syncd /tmp
hd1 jfs2 1 1 1 open/syncd /home
hd10opt j fs2 1 1 1 open/syncd /opt
lg_dumplv sysdump 80 80 1 open/syncd N/A
download jfs2 200 200 1 open/syncd /downloads
4. Run the mirrorvg command to create the relationship and start the copy of the data:
root:/ > mirrorvg rootvg hdisk1
After the mirroring process ends, we have the following output for the lsvg -l command.
Now you will see that the proportion between LPs and PPs columns is 1 by 2 that means 1
logical data in 2 physical volumes:
root:/ > lsvg -l rootvg
rootvg:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
hd5 boot 1 2 2 closed/syncd N/A
hd6 paging 8 16 2 open/syncd N/A
hd8 jfs2log 1 2 2 open/syncd N/A
hd4 jfs2 1 2 2 open/syncd /
hd2 jfs2 21 42 2 open/syncd /usr
hd9var jfs2 1 2 2 open/syncd /var
hd3 jfs2 1 2 2 open/syncd /tmp
hd1 jfs2 1 2 2 open/syncd /home
hd10opt jfs2 1 2 2 open/syncd /opt
lg_dumplv sysdump 80 160 1 open/syncd N/A
download jfs2 200 400 2 open/syncd /downloads
Now the volume group is mirrored and the data is consistent in the 2 volumes as shown in
the columns LV STATE that is indicating the status syncd for all logical volumes.
If you want to remove the mirror, you can use the following command:
#unmirrorvg <vg_name> <hdisk#>
If you want to remove the hdisk1 and keep the hdisk0 active, run the following command:
#unmirrorvg rootvg hdisk1
If you want to remove the hdisk0 and keep the hdisk1 active, run the following command:
#unmirrorvg rootvg hdisk0
Note: You can use smit utility to perform these procedures accessing the fast path smit
mirrorvg to create a mirror or smit unmirrovg to remove a mirror.
In this example we have two volumes, Disk 8 and Disk 9. The drive letter S: is associated with
Disk 8, which is the current volume running on the system. Disk 9 is the new disk that will be
part of the mirror. See Figure A-1.
Note: To configure new disks on the system, after connecting it to the Windows server, run
the Rescan Disks function in the Disk Management.
Figure A-2 shows how to convert the new volume to dynamic disk.
Now, with Disk 9 as a dynamic disk, the system is ready to initiate the mirroring process. As
Figure A-3 illustrates, right-click the source volume (S:) and choose the Add Mirror option.
The Add Mirror window will be displayed with a list of available disks. Mark the chosen disk
and then click Add Mirror; see Figure A-4.
The synchronization process will start automatically. At this time you are will see that both
volumes, Disk 8 and Disk 9, are assigned to the same drive letter S:; see Figure A-5.
Figure A-6 shows the volumes after the synchronization process has finished.
The next panels show you how to remove a mirror. We can access this option by
right-clicking the selected volume. You have two options now:
Break Mirrored Volume
The selected volume will keep the original drive letter and the other volume will be
automatically assigned to another letter. From this time the synchronization process will
not occur; both drives will have different drive letters, but the data is still on it.
Remove mirror
If you choose to remove the mirror, a window will be displayed asking you which volume
you want to remove. The selected volume, after completing the process, will become a
free disk to the operating system with no drive letter and no data inside.
In Figure A-7 we select the option Break Mirrored Volume from Disk 8.
After you confirm the operation, you will see that Disk 8 changes to drive letter E:; see
Figure A-8. The data is still available, but the disks will not be fault-tolerant.
Figure A-9 shows you the disks with the mirror, and this time we selected the Remove Mirror
option. A window opens in which you select which disk will be removed. We select Disk 8.
After selecting Remove Mirror, the selected volume will became available to the operating
system without a drive letter and no data available. See Figure A-10.
All open system platforms and many applications provide native backup and restore
capabilities. They may not be very sophisticated sometimes, but they are often suitable in
smaller environments. In large data centers it is customary to have a common backup
solution across all systems. Either can be used for data migration.
The backup and restore options allow for consolidation because the tools are aware of the
data structures they handle.
One significant difference to most of the other methods is that this method may not require
the source and target storage systems to be connected to the hosts at the same time.
Some of the most common software packages that provide this facility are:
IBM Tivoli Storage Manager (TSM)
Legato Networker
BrightStor ARCserve
Veritas NetBackup
This section discusses methods for migrating data from existing disk storage systems onto
the DS8000. Our intention here is to show the possibilities that you have and not provide a
detailed step-by-step migration process description. The following topics are discussed:
Data migration based on physical volume migration
Data migration based on logical data set migration
Combination of physical and logical data migration
Copy Services based migration
The following software products and components can be used for data logical migrations:
DFSMS allocation management.
Allocation management by CA-ALLOC.
DFSMSdss.
DFSMShsm.
FDR.
System utilities like:
– IDCAMS with REPRO, EXPORT/IMPORT commands
– IEBCOPY for Partitioned Data Sets (PDS) or Partitioned Data Sets Extended (PDSE)
– ICEGENER as part of DFSORT, which can handle sequential data but not VSAM data
sets, which also applies to IEBGENER
CA-Favor.
CA-DISK or ASM2.
Database utilities for data that is managed by certain database managers like DB2 or IMS.
CICS as a transaction manager usually uses VSAM data sets.
Then perform the logical data set copy operation to the larger volumes. This allows you to
use either DFSMSdss logical copy operations or the system-managed data approach.
When a level is reached where no data moves anymore because the remaining data sets are
in use all the time, some down time has to be scheduled to perform the movement of the
remaining data. This might require you to run DFSMSdss jobs from a system that has no
active allocations on the volumes that need to be emptied.
For additional information about available migration services you can refer to the following
IBM Web site: http://www-1.ibm.com/servers/storage/services/disk.html.
Summary
This appendix shows that there are many ways to accomplish data migration. Thorough
analysis of the current environment, evaluation of the requirements, and planning are
necessary. Once you decide on one or more migration methods, refer to the documentation of
the tools you want to use to define the exact sequence of steps to take. Special care must be
exercised when data is shared between more than one host.
The migration might be used as an opportunity to consolidate volumes at the same time.
Capacity Magic
Because of the additional flexibility and configuration options they provide, it becomes a
challenge to calculate the raw and net storage capacity of disk subsystems like the DS8000
and the DS6000. The user has to put some considerable time and needs an in-depth
technical understanding of how spare and parity disks are assigned, as well as take into
consideration the simultaneous use of disks with different capacities and configurations that
deploy both RAID 5 and RAID 10.
Capacity Magic, a product from IntelliMagic, is there to do the physical (raw) to effective (net)
capacity calculations automatically, taking into consideration all applicable rules and the
provided hardware configuration —number and type of disk drive sets.
Capacity Magic is designed as an easy-to-use tool with a single main dialog. It offers a
graphical interface that allows you to enter the disk drive configuration of a DS8000, DS6000,
or ESS 800; the number and type of disk drive sets; and the RAID type. With this input
Capacity Magic calculates the raw and net storage capacities —also some new functionality
has been introduced into the tool to display the number of extents that are produced per rank.
Figure B-1 shows the configuration screen that Capacity Magic provides for you to specify the
desired number and type of disk drive sets.
Figure B-2 on page 421 shows the resulting output report that Capacity Magic produces. This
report is also helpful to plan and prepare the configuration of the storage in the DS8000, as it
also displays extent count information.
Note: Capacity Magic is a product of IntelliMagic and is protected by a license code. For
more information, visit the following Web site: http://www.intellimagic.net/en/.
Disk Magic
Disk Magic, a product from IntelliMagic, is a Windows-based disk subsystem performance
modeling tool. It supports disk subsystems from multiple vendors, but it offers the most
detailed support for IBM subsystems.
The first release was issued as an OS/2 application in 1994, and since then Disk Magic has
evolved from supporting storage control units such as the IBM 3880 and 3990 to supporting
modern, integrated, advanced-function disk subsystems such as the DS8000, DS6000, ESS,
DS4000, and the SAN Volume Controller.
A critical design objective for Disk Magic is to minimize the amount of input the user must
enter, while offering a rich and meaningful modeling capability. The following list provides
some examples of what Disk Magic can model, but it is by no means complete:
Move the current I/O load to a different disk subsystem model
Merge the I/O load of multiple disk subsystems into a single one
Insert a SAN Volume Controller in an existing disk configuration
Increase the current I/O load
Implement a storage consolidation
Increase the disk subsystem cache size
Change to larger capacity disk drives
Upgrade from SCSI to Fibre Channel host adapters
Use fewer or more Logical Unit Numbers (LUN)
Activate Metro Mirror
Figure B-3 shows some of the panels that Disk Magic provides for you to input data —the
example shows the panels for input of host adapters information and disk configuration in an
hypothetical case of a DS8000 attaching open systems. Modeling results are presented
through tabular reports.
Figure B-3 Disk Magic Interfaces and Open Disk input screens
Note: Disk Magic is a product of IntelliMagic and is protected by a license code. For more
information, visit the following Web site: http://www.intellimagic.net/en/.
IBM TotalStorage Productivity Center for Disk centralizes management of networked storage
devices that implement the SNIA SMI-S specification, which includes the IBM System
Storage DS family and SAN Volume Controller (SVC). It is designed to help reduce storage
management complexity and costs while improving data availability, centralizing management
of storage devices through open standards (SMI-S), enhancing storage administrator
productivity, increasing storage resource utilization, and offering proactive management of
storage devices. IBM TotalStorage Productivity Center for Disk offers the ability to discover
storage devices using Service Location Protocol (SLP) and provides the ability to configure
devices, in addition to gathering event and errors logs and launching device-specific
applications or elements.
For more information see Managing Disk Subsystems using IBM TotalStorage Productivity
Center, SG24-7097. Also, refer to:
http://www.ibm.com/servers/storage/software/center/index.html.
Within this service, IBM will propose a possible configuration for the target disk subsystem
that is based on the information the user provides in a questionnaire and the configuration of
the currently used storage disk systems.
Linux or Windows
ThinkPad with
‘Disk Storage Configuration
Migrator’
Capture current
configuration(s) to
generate new DS8000
configuration(s)
The standard CLI interfaces of the ESS and the DS8000 are used to read, modify, and write
the logical and Copy Services configuration. All information is saved in a data set in the
provided database on a workstation. Via the Graphical User Interface (GUI), the user
information gets merged with the hardware information and it is then applied to the DS8000
subsystem. See Figure B-4.
Note: This approach could also be used to convert DS4000, HSG80, EMC, and Hitachi to
DS8000.
For more information, contact your IBM representative or IBM Business Partner, or visit
http://ibm.com/services/
For details on available services, contact your IBM representative or visit the following:
http://www.ibm.com/services/
http://www-03.ibm.com/servers/storage/services/disk.html
For details on available IBM Business Continuity and Recovery Services, contact your IBM
representative or visit http://www.ibm.com/services/continuity
Select your country, and then select the product as the category.
Related publications
The publications listed in this section are considered particularly suitable for a more detailed
discussion of the topics covered in this redbook.
IBM Redbooks
For information about ordering these publications, see “How to get IBM Redbooks” on
page 431. Note that some of the documents referenced here may be available in softcopy
only.
IBM System Storage DS8000 Series: Copy Services with System z servers, SG24-6787
IBM System Storage DS8000 Series: Copy Services in Open Environments, SG24-6788
IBM TotalStorage Business Continuity Solutions Guide, SG24-6547
The IBM TotalStorage Solutions Handbook, SG24-5250
IBM TotalStorage Productivity Center V2.3: Getting Started, SG24-6490
Managing Disk Subsystems using IBM TotalStorage Productivity Center, SG24-7097
IBM TotalStorage: Integration of the SAN Volume Controller, SAN Integration Server, and
SAN File System, SG24-6097
IBM TotalStorage: Introducing the SAN File System, SG24-7057
If you are implementing Copy Services in a mixed technology environment you may be
interested in referring to the following redbooks on the ESS and the DS6000.
IBM TotalStorage Enterprise Storage Server Implementing ESS Copy Services in Open
Environments, SG24-5757
IBM TotalStorage Enterprise Storage Server Implementing ESS Copy Services with
IBM Eserver zSeries, SG24-5680
IBM System Storage DS6000 Series: Copy Services with System z servers, SG24-6782
IBM System Storage DS6000 Series: Copy Services in Open Environments, SG24-6783
DFSMShsm ABARS and Mainstar Solutions, SG24-5089
Practical Guide for SAN with pSeries, SG24-6050
Fault Tolerant Storage - Multipathing and Clustering Solutions for Open Systems for the
IBM ESS, SG24-6295
Implementing Linux with IBM Disk Storage, SG24-6261
Linux with zSeries and ESS: Essentials, SG24-7025
Other publications
These publications are also relevant as further information sources. Note that some of the
documents referenced here may be available in softcopy only.
IBM System Storage DS8000: Command-Line Interface User´s Guide, SC26-7916
IBM System Storage DS8000: Host Systems Attachment Guide, SC26-7917
Online resources
These Web sites and URLs are also relevant as further information sources:
IBM System Storage Ds8000 Information Center
http://publib.boulder.ibm.com/infocenter/ds8000ic/index.isp
IBM Disk Storage Feature Activation (DSFA) Web site
http://www.ibm.com/storage/dsfa
The PSP information
http://www-1.ibm.com/servers/resourcelink/svc03100.nsf?OpenDatabase
Documentation for the DS8000
http://www.ibm.com/servers/storage/support/disk/2107.html
The Interoperability Matrix
http://www-1.ibm.com/servers/storage/disk/ds8000/interop.html
Fibre Channel host bus adapter firmware and driver level matrix
http://knowledge.storage.ibm.com/servers/storage/support/hbasearch/interop/hbaSearch.do
ATTO
http://www.attotech.com/
Emulex
http://www.emulex.com/ts/dds.html
JNI
http://www.jni.com/OEM/oem.cfm?ID=4
QLogic
http://www.qlogic.com/support/ibm_page.html
IBM
http://www.ibm.com/storage/ibmsan/products/sanfabric.html
McDATA
http://www.mcdata.com/ibm/
Cisco
http://www.cisco.com/go/ibm/storage
CIENA
http://www.ciena.com/products/transport/shorthaul/cn2000/index.asp
CNT
http://www.cnt.com/ibm/
Nortel
http://www.nortelnetworks.com/
ADVA
http://www.advaoptical.com/
Index
Control Unit Initiated Reconfiguration see CUIR
A cooling 86
AAL 54–55 disk enclosure 66
benefits 55 rack cooling fans 87
activate licenses Copy Services
applying activation codes using the DS CLI 209 event traps 386
functions 198 interfaces 124
address groups 101 CUIR 82
Advanced Function Licenses 155
activation 155
AIX 299 D
boot support 307 DA 50
I/O access methods 306 Fibre Channel 170
LVM configuration 305 daemon 382
on iSeries 370 data migration 415
WWPN 299 backup and restore 415
AIX MPIO 302 Disk Magic 421
applying activation codes using the DS CLI 209 Disk Storage Configuration Migrator 423
architecture 42 IBM Migration Services 417
array sites 94 logical migration 416
arrays 54, 94 open systems environments 406
arrays across loops see AAL physical and logical data migration 416
physical migration 415
summary 417
B volume management software 407
backup and restore 415 zSeries environments 415
base frame 40 data placement 105
battery backup assemblies 66 Data Set FlashCopy 112
battery backup unit see BBU DDMs 55
BBU 87 hot plugable 86
boot support 290 device adapter see DA
Business Continuity 4 disk
business continuity 8 raw 323
disk drive set 7
C disk drives
cables 138 capacity 143
cache management 44 FATA 56, 177
Call Home 385 disk enclosure 50
Call Home/call back 392 power and cooling 66
Capacity Magic 420 Disk Magic 421
CCL Disk Storage Configuration Migrator 423
Command Console LUN (CCL) see CCL disk subsystem 50
CKD volumes 98 disk virtualization 93
allocation and deletion 99 DS API 108
community name 384 DS CLI 108, 126, 140
components 39 applying activaton codes 209
used in the DS HMC environment 147 configuring second HMC 163
configuration console
FC port 316 user management 159
volume 317 DS Command-Line Interface see DS CLI
configuration flow 220 DS HMC
configuring the DS8000 271, 280 Call Home/call back 392
Configuring using DS CLI external 162
configuring the DS8000 271, 280 DS HMC planning 145
Consistency Group FlashCopy 113 activation of Advanced Function Licenses 155
Index 435
6786IX.fm Draft Document for Review November 14, 2006 3:49 pm
Index 437
6786IX.fm Draft Document for Review November 14, 2006 3:49 pm
Remote Mirror and Copy 36, 108, 142 SNMP agent 382–383
ESCON 64 SNMP manage 382
Remote Mirror and Copy function see RMC SNMP trap 382, 384
Remote Mirror and Copy see RMC SNMP trap request 382
remote power control 141 software
remote support 140 setup 154
Requests for Price Quotation see RPQ spares 54, 85
RIO-G 48 floating 85
RMC 9, 115 sparing considerations 142
Global Copy 117 SPCN 48
Global Mirror 117 Standby Capacity on Demand see Standby CoD
Metro Mirror 116 Standby CoD 7
room space 135 storage area network connection 142
RPC 65 storage capacity 7
RPO 124 storage complex 70
RPQ 290 storage facility image 30, 70
addressing capabilities 37
hardware components 32
S I/O resources 32
SAN 81 processor and memory allocations 32
SAN LUNs 370 RIO-G interconnect separation 32–33
SARC 12, 44 Storage Hardware Management Console see S-HMC
SATA 56 storage unit 70
scalability 10 stripe
DS8000 size 178
scalability 174 summary 417
SDD 12, 179, 300 switched FC-AL 6
for Windows 292 advantages 52
Sequential Prefetching in Adaptive Replacement Cache DS8000 implementation 53
see SARC System i
serial number 155 protected volume 345
server system power control network see SPCN
RAS 75
server-based SMP 44
service clearance 135 T
service processor 48 time synchronization 157
settings tools
HBA and operating system 291 Capacity Magic 420
S-HMC 7, 66, 89, 124 TPC for Disk 422
simplified LUN masking 10 trap 382, 384
sizing troubleshooting and monitoring 314
open systems 176
z/OS 180
SMUX 382 U
SNMP 382 user managemen using DS SM 161
configuration 385, 390 user management
Copy Services event traps 386 using DS CLI 159
notifications 385 using DS SM 161
preparation for the management software 390
preparation on the DS HMC 390 V
preparation with DS CLI 390 VDS support
trap 101 386 Windows Server 2003 295
trap 202 388 virtualization
trap 210 388 abstraction layers for disk 93
trap 211 388 address groups 101
trap 212 388 array sites 94
trap 213 389 arrays 94
trap 214 389 benefits 106
trap 215 389 concepts 91
trap 216 389 definition 92
trap 217 389
Index 439
6786IX.fm Draft Document for Review November 14, 2006 3:49 pm
extent pools 96
hierarchy 104
host attachment 102
logical volumes 97
ranks 95
storage system 92
volume group 103
VMFS 323
volume groups 103
volumes
CKD 98
VSE/ESA 341
W
Windows 291
SDD 292
WWPN 299
Z
z/OS
considerations 335
VSE/ESA 341
z/OS Global Mirror 9, 108, 122, 124
z/OS Metro/Global Mirror 9, 119, 122
z/VM considerations 340
zSeries 415
host connection 82
host considerations 334
performance 13
prerequisites and enhancements 334