You are on page 1of 56

® ®

Oracle SAP

for
T E C H N O L O G Y U P D A T E

g
10
D A T A B A S E

Oracle for SAP


is safe, reliable
and scalable
– the Grid enabled
platform

No. 15 Oracle for SAP, April 2006


www.oracle.com/newsletters/sap
Oracle for SAP
is safe, reliable and scalable
– the Grid enabled platform

10 g
D A T A B A S E

CONTENTS

1 Editorial
2 Update: Oracle Database 10g and Real Application Clusters 10g for mySAP
6 SAP on Real Application Clusters (RAC) at Stadtwerke Munich (Windows)
8 Oracle Real Application Clusters (RAC) Advantages for SAP Customers
10 SAP on Real Application Clusters (RAC) at S.R.I., Fujitsu Siemens, (Solaris)
12 SAP on Real Application Clusters (RAC) at Gas Natural, BW (Linux)
15 SAP on Real Application Clusters (RAC) Certified Configurations for Windows and Linux
16 Dr. Oetker: DB2 to Oracle migration to SAP BW
18 Informix: Oracle database migration at Testo
22 Oracle Data Mining for SAP Business Warehouse Customers
25 Delta Consulting: RAC Services for SAP Customers
26 ChevronTexaco: Oracle Recovery Manager (RMAN) for SAP at ChevronTexaco
28 Burda: HP Tru64 to HP-UX Oracle to Oracle Migration
31 COOP Switzerland: Oracle Advanced Customer Services
34 Japan Aviation Electronics (JAE) using Oracle Customer Services for SAP
36 Dell: Configuring a Highly Available Linux Cluster for SAP Services
44 Oracle RAC and Egenera
46 SAP INFO Solutions: Active Global Support – What’s New?
47 TUSC: Oracle Partner for SAP
48 Oracle PREMIUM Services for SAP R/3 Customersr s
51 BPEL: Integration to SAP, Webservices
54 Useful Links Oracle for SAP Customers
54 Imprint
Oracle for SAP 1

DEAR SAP CUSTOMER,

I hope you will enjoy reading this 15th Edition of the Oracle for SAP Global
Technology Update.
For over 17 years Oracle has been the leading database for SAP customers and we
remain committed to our long term goal to provide safe, reliable, and scalable
database technology for our SAP customers. Two thirds of SAP customers run their
SAP application on an Oracle database, leveraging their investment in Oracle technology.
As an ongoing commitment to our joint customer base, SAP and Oracle have extended
the long standing Development, Reseller and Support relationship.

Oracle9i Real Application Clusters (RAC) is now available on all HW platforms and Oracle 10g Release 2 is now
shipping to SAP customers! This is a major milestone for SAP customers running the Oracle database; it enhances
database availability, scalability, and it lowers the TCO. While unprecedented market pressure to decrease costs
and provide fast interconnect technology prefaced the development of grid technology, the development of
Oracle 10g grid technology will provide SAP customers with additional benefits.

In this edition of the Oracle for SAP Technology Update you will read about Oracle’s 10g technology benefits
to SAP customers and Oracle Data Mining Technology integration, which provides SAP customers advanced
analytics embedded in SAP BW as an optional feature. We have also included a significant number of new
SAP on Oracle Real Application Clusters (RAC) customer success stories.

You will read about Oracle Services and Support available to SAP customers including; database migrations,
and performance tuning. Also learn how customers worldwide have reduced costs by migrating their Informix
database under SAP to Oracle. The Oracle database technology has a proven track record to minimize IT costs
throughout the whole life cycle of an SAP application.

We hope you enjoy this edition of the newsletter!

Any news about this and other topics will be published immediately on our web site, so check back periodically
at: www.oracle.com/newsletters/sap Please don't hesitate to contact us with your questions or comments.
E-mail: saponoracle_de@oracle.com For price information about Oracle9i Real Application Clusters contact your
Oracle or SAP Account Manager or E-mail us at: saponoracle_de@oracle.com

Sincerely

Gerhard Kuppler

Senior Director Corporate SAP Account


Oracle Corporation
2

U P D AT E – O R A C L E D AT A B A S E 1 0g A N D
R E A L A P P L I C AT I O N C L U S T E R S 1 0g F O R M Y S A P

INTRODUCTION Online Segment Shrink

Oracle Database 10g has been designed to reduce the This feature shrinks segments of tables and indexes
g
10 cost of manageability, to deliver increased performance online and in-place that have free space in them,
for all types of workloads and to provide major new thereby enhancing efficiency of space utilization.
high-availability features.
D A T A B A S E Automatic Optimizer Statistics Collection
Oracle Database 10g is the first database designed for This feature automates the collection of optimizer
Enterprise Grid Computing to lower hardware costs statistics for objects. Objects with stale or no statistics
by leveraging low cost components and by greatly are automatically analyzed, so administrators no
increasing resource utilization levels, therefore com- longer need to keep track of what does and what does
plementing the Advanced Computing Infrastructure not need to be analyzed, nor to perform analysis by
concept of SAP. hand.
As with previous Oracle database versions, all trans-
parent features of Oracle Database 10g are immediately End-to-End Application Tracing
available for all types of SAP applications. The initial The feature simplifies the debugging of performance
certification of Oracle Database 10g by SAP has been problems in multi-tier environments. It will only be
done at the end of the first quarter of 2006. After the available for new versions of the SAP software and
initial certification more new Oracle Database 10g not for existing versions.
features will be adopted by SAP over time.
Online LONG to LOB Migration
F E AT U R E S I M M E D I AT E LY In this release tables with LONG and LONG RAW
SUPPORTED BY SAP data, still used by many SAP applications, can now
be migrated online to LOB data from within the
The features described in this section are available
BR*SPACE utility of SAP.
for and offer benefits to all types of SAP applications.
They are supported with the initial certification of
NEW FUNCTIONALITY WITH
Oracle Database 10g now.
SAP BW 7.0
Instant Client SAP BW users will benefit most from the new features
Instant Client allows you to run your applications describes in this section. They are supported with
without installing the standard Oracle client or the initial certification of Oracle Database 10g now.
having an ORACLE_HOME. OCI-based applications
like mySAP work without modification, while using Faster Drop and Truncate Table
significantly less disk space than before. These two operations perform substantially faster by
using improved algorithms in accessing the database
Rename Tablespace buffer cache. In particular small tables, often used by
You can now rename a tablespace. You no longer have SAP BW, benefit the most.
to create a new tablespace, copy the contents from the
old tablespace, and drop the old tablespace. This
feature makes it easier, for example, to migrate a dic
to transport a tablespace to a database that already
contains a tablespace of the same name.
Oracle 10g Update for SAP – supported features 3

In Memory Undo Integrated Clusterware


Block changes by short transactions are now managed Oracle RAC 10g offers a complete clusterware mana-
cleverer by the database server resulting in less CPU gement solution as an integral component of Oracle
cycles. RAC 10g available on all platforms Oracle Database
10g runs on. This clusterware functionality includes
Improved Scalability for Partitioned Objects mechanisms for cluster connectivity, messaging and
Dropping partitioned tables and indexes is now locking, cluster control and recovery, and a services
substantially faster by using a new algorithm for provisioning framework. No 3rd party clusterware
identifying the blocks of a partitioned object in the management software need be purchased. Oracle will,
buffer cache and for cleaning these blocks in the however, continue to support select 3rd party cluster-
buffer cache. The SAP BW application benefits the ware products on specified platforms.
most, as dropping partitioned objects is a frequently With Oracle Database 10g Release 2, Oracle provides
used operation in SAP BW. a High Availability API so that non-Oracle processes
Tests performed using a query on a table with 1,000 can be put under the control of the high availability
partitions show that Oracle 10g’s new strategy of framework within Oracle Clusterware.
sharing partitioning metadata results in less SQL mem-
Single System Image Management
ory usage in the SGA (9.2.0.6: 52KB, 10.2: 11KB –
this means that every single query on this table will Oracle Enterprise Manager 10g has been significantly
use 5 times less memory). enhanced to enable true single system image manage-
ment of cluster database deployments. Enterprise
Manager’s Cluster Database Page provides a single
O R A C L E R E A L A P P L I C AT I O N view of system status across multiple nodes. It also
CLUSTERS 10g enables direct drill down to individual instances when
Oracle Real Application Clusters (RAC) allows Oracle needed.
Database to run any packaged or custom application,
unchanged across a set of clustered servers. This pro- A new cluster configuration verification tool and
vides the highest levels of availability and the most improvements in the diagnostic tools first introduced
flexible scalability. To keep costs low, even the high- in the Oracle9i Database release help users both avoid
est-end systems can be built out of standardized, com- problems and resolve problems more quickly should
modity parts. Oracle 10g RAC provides a foundation they occur.
for Oracle’s Enterprise Grid Computing Architecture.
Data Guard Integration for Disaster Recovery
Oracle RAC technology enables a low-cost hardware
platform to deliver the highest quality of service that With Oracle Enterprise Manager 10g, the manage-
rivals and exceeds the levels of availability and scala- ment component of Oracle Data Guard, Data Guard
bility achieved by the most expensive, mainframe Broker, is now completely integrated with RAC.
SMP computers. Data Guard disaster recovery environments involving
Oracle RAC databases can now be as easily managed
Oracle 10g RAC is not included in the initial certifi- as those employing single instance databases.
cation of Oracle Database 10g. In the same way as for
Oracle 9.2 RAC, a separate release process is carried F E AT U R E S T O B E S U P P O R T E D
out. During the second and third quarters of 2006,
L AT E R B Y S A P
there are one or two Oracle 10g pilot projects planned
that will lead to Oracle 10g RAC being released in Backup Compression
the SAP environment during the third quarter of If disk space is an issue, or your media-management
2006. software does not support compression, RMAN pro-
vides the ability to compress RMAN backup sets.
4

Full Database Begin Backup Command Flashback Technology


It is no longer necessary to issue a separate command In this release, Oracle introduces expanded database
to place each tablespace in hot backup mode. You flashback capabilities. If a major error occurs, such as
can now use the ALTER DATABASE statement to a batch job being run twice in succession, the data-
place all tablespaces in backup mode. Also, the base administrator can request a flashback operation
BEGIN BACKUP command now runs faster than that quickly recovers the entire database to a previous
before. point in time, eliminating the need to restore back-
ups and do a point-in-time recovery. In addition to
Change-Aware Incremental Backup flashback operations at the database level, it is also
By using a new type of log file to track blocks that possible to flash back an entire table in this release
have changed in the database, RMAN can avoid of the Oracle database. Similarly, a new capability
scanning the entire datafile during an incremental allows the database to recover tables that have been
backup. Instead, the amount of data scanned is pro- inadvertently dropped by a user. The existing flash-
portional to the amount of data changed. back query capabilities have also been improved.

Secured Redo Transmission • Flashback Database: This feature introduces the


When using the Advanced Security Option of Oracle FLASHBACK DATABASE statement in SQL. It
Database 10g the security of a Data Guard environ- let you quickly bring your database to a prior point
ment can be increased by preventing possible tampe- in time by undoing all the changes that have taken
ring of redo data as it is being transferred to the place since that time. This operation is fast, because
standby database. As long as SAP has not certified you do not need to restore the backups. This in turn
the Advanced Security Option this feature will not results in much less downtime following data
be available for SAP installations. corruption or human error.
Data Pump Export and Import Utilities • Flashback Reinstantiation: This feature reduces
The Data Pump Export and Import utilities provide the need to reinstantiate the old primary database
very high-speed bulk movement of data and meta- following a failover. This in turn lets you more
data from database to another. These utilities offer quickly restore full resiliency after a failure. This
several significant advantages over the original is done by using the SQL statement FLASHBACK
Export and Import utilities, including: the ability to DATABASE to roll back the primary database in
completely restart export and import jobs; the ability time to synchronize with the standby database.
to detach from and reattach to long-running jobs;
the ability to estimate how much space an export job • Flashback Standby Database: This feature
would consume; support for export and import ope- improves the switchover and failover time of a
rations over the network; and support for fine-grained standby database. You no longer need to specify a
object selection, based upon objects and objects types. log apply delay, because you can now roll back the
This feature will initially not be integrated within standby database if some error occurs on the primary
the SAP administration tools. and is propagated to the standby.

The new Data Pump Export and Import utilities can • Flashback Table: This feature introduces the
each be run in parallel, resulting in better performance FLASHBACK TABLE statement in SQL, which
in loading and unloading data and metadata. lets you quickly recover a table to a point in time
in the past without restoring a backup.
Cross-Platform Transportable Tablespaces
The transportable tablespace feature now enables • Flashback Versions Query: Using undo data
tablespaces to be transported across different plat- stored in the database, you can now view the
forms therefore making a platform migration for a changes to one or more rows along with all the
SAP customer much faster. metadata of the changes.
Oracle 10g Update for SAP – supported features 5

• Flashback Drop (Recycle Bin): Oracle now A D D I T I O N A L I N F O R M AT I O N


provides a way to restore accidentally dropped
The SAP Database Guide: Oracle, which is available
tables. Prior to Oracle 10g, a DROP command
via the Media Library on service.sap.com, describes
permanently removed objects from the database.
support for new Oracle 10g features in SAP NetWeaver
In Oracle 10g, a DROP command places the object
2004s. The Database Upgrade Guide (Upgrade to Oracle
in the recycle bin. The extents allocated to the
Database 10g) describes the technical details of the
segment are not reallocated until you purge the
upgrade process.
object. You can restore the object from the recycle
bin at any time. Within SQL*PLUS you are SAP OSS Note 720886 is the master entry point for
able through the SHOW RECYCLEBIN [origi- information about Oracle Database 10g integration
nal_name] command to view the objects that are into the SAP environment. OSS Note 828268 provides
available for purging or reverting using the PURGE an overview of new features that can be used in the
and FLASHBACK BEFORE DROP commands. SAP environment and points to other OSS notes that
discuss individual features in more detail (see page 54).
• Flashback Transaction Query: This feature intro-
duces flashback transaction query, which lets you
CONCLUSION
examine changes to the database at the transaction
level. As a result, you can diagnose problems, Oracle Database 10g provides the necessary infra-
perform analysis, and audit transactions. structure of making the SAP landscape more respon-
sive to the changing business environment. Oracle
Automated Storage Management (ASM) Database 10g gives you the flexibility you need to
ASM automates and simplifies the optimal layout of implement Enterprise Grid Computing, even using
datafiles, control files, and log files. Database files are low cost servers and disks. The design of Enterprise
automatically distributed across all available disks, and Grid Computing complements the Adaptive
database storage is rebalanced whenever the storage Computing Infrastructure concepts of SAP. Oracle
configuration changes. This feature also provides Database 10g provides the benefit of lowering risks
redundancy through the mirroring of database files. with lower manageability costs, greater scalability
ASM was particularly designed for using low cost and predictability and the highest levels of availa-
servers and disks. However, ASM support by SAP bility.
will not be available with the initial certification of
Oracle Database 10g though. ASM will be added
later only for installations of new versions of SAP
products. A migration path for older versions of SAP
products will follow once the initial certification for
ASM is been done by SAP.

F E AT U R E S T O B E S U P P O R T E D
L AT E R B Y S A P
The SAP Database Guide: Oracle, which is available
via the Media Library on service.sap.com, describes
support for new Oracle 10g features in SAP
NetWeaver 2004s. The Database Upgrade Guide
(Upgrade to Oracle Database 10g) describes the tech-
nical details of the upgrade process.
SAP OSS Note 720886 is the master entry point for
information about Oracle Database 10g integration
into the SAP environment. OSS Note 828268 provides
an overview of new features that can be used in the
SAP environment and points to other OSS notes
6

“The objectives which we had pursued with migration towards Oracle Real Application
Clusters have been greatly surpassed. The improvements with respect to costs, availability,
performance and scalability can be clearly evidenced. We have been able to optimize SAP
use effectively in Microsoft Windows with Oracle RAC and are therefore in a strong position
for the future”

— RUDOLF BAYERL
Head of System Operations, Area of SWM Services GmbH
Information and Process Technology

S TA D T W E R K E M U N I C H I S P U R S U I N G N E W AV E N U E S
FOR SAP USE IN WINDOWS WITH ORACLE REAL
A P P L I C AT I O N C L U S T E R S ( R A C )

Intensive use of SAP entails higher ting, purchase handling, maintenance manage-
requirements ment and much, much more. In total, SWM
has around 3000 users. The Microsoft Windows
The Stadtwerke Munich (SWM) is the largest
platform combined with an Oracle database is
municipal enterprise in Germany with around
Stadtwerke Munich the preferred medium for SAP use at the end
7000 employees and group revenue of 2.95
Services GmbH of the mainframe era, though open-source
billion euros (2004). From the private indivi-
platforms (above all Linux) have also figured
Industry: dual or medium-sized company through to the
in the considerations for quite a while now.
Utility/Public Sector global player, customers of all types can find a
These are also based on Oracle, and more-over:
Annual Revenue: partner in SWM to supply them with power,
the extent of SAP use has been – and still is –
water and associated services reliably and cost
2.95 billion euros (2004) continuously increasing. The SAP database
effectively. In addition to power supply and
Employees: volume is growing by around 50 gigabytes a
services, the spectrum offered encompasses the
Around 7,000 month, and currently totals around 2 terabytes.
complete energy and water value-added chain.
Oracle Products & Services:
In addition to SAP, SWM has well over 100
SWM is also a municipal service provider
other applications which operate together on
• Oracle database 9.2.0.5 offering high-quality service and promoting
an Oracle database. Extensive experience has
• Oracle Real Application leisure activities in the world famous city on
Clusters (RAC)
already been gathered in non-SAP environ-
the Isar, thanks to its indoor and outdoor
• Oracle OCFS ments with the innovative database clustering
swimming pools. And with the comprehensive
• Oracle Support technology Oracle Real Application Clusters.
mobility package of MVG (subway, bus and
Key Benefits tram), SWM ensures the inhabitants of Munich
Good experience with Oracle RAC in a
• Options for optimizing cost mobility in a way that is friendly to the city
non-SAP environment
efficiency and environment. To provide its services,
• Higher failure reliability SWM is available around the clock – 24 hours “Ever more demanding challenges for SAP
• Improved scaling potentials a day, 365 days of the year. applications, combined with a growing data
• Performance improvements volume and increasingly deteriorating perfor-
• Increase in investment mance, were the reasons behind the reorienta-
This type of “all-round supply” has important
safeguarding
prerequisites for the IT systems used. These tion towards RAC. We also wanted to utilize
too have to be available around the clock – the advantages which Oracle RAC provides for
Improvement of especially the SAP solutions used. SAP R/3 SAP use in Windows, especially as our existing
High Availability (Version 4.6C) and the branch solution IS- experience was very positive”, declares Rudolf
Cost Reduction Bayerl, Head of System Operations in the area
U/CCS (Version 4.64) enable the management
Long-term IT concept
of customer and supplier data, service accoun- of Information and Process Technology at SWM
to ensure ROI
Streamlined DBA
productivity
State-of-the-art High-
Availability Concept
S A P R eOarlaA
c lpep D
lia
cat at iboans eC l1u0sgt e–r sI nafto rSm
t aadt itowne ri sk eC u
Mrur enni cchy 7

Services GmbH. Oracle RAC has enabled the


longstanding SAP customer SWM (which had
already employed SAP R/2 before SAP R/3) to
combine several advantages: hardware invest-
ment safeguarding, increase in failure reliabi-
lity, performance increases and additional
options for improved scalability.

After several discussions with Oracle, Microsoft


Cluster File System (OCFS) over several weeks
and SAP as well as the storage supplier EMC
as part of the tuning measures. We achieved
and the hardware provider Dell, a project was
our mutual goal on a step by step basis. The
formulated with the following objective:
commitment shown by the partners involved
Oracle RAC was to be used together with SAP
was remarkable. The entire project team set
R/3/IS-U/CCS and Microsoft Windows 2003
about its tasks with dedicated enthusiasm,”
Server Edition as a productive application at
emphasizes Manfred Fischer, Systems Engineer
SWM for the first time in the world as part of
at SWM Services GmbH.
a pilot project. April 1st 2004 was earmarked
as the project start. Existing Microsoft clusters
were to be used for the central instance, while Surpassing objectives previously set
an RAC configuration was to be employed for SAP productive operation with Oracle Real
the database. In the case of SWM, this meant Application Clusters on the operating system
that Oracle database clustering was to be platform Windows 2003 Server Edition and
operated on 3 nodes with three instances from clustered Dell computers commenced at the
then on, instead of a conventional central beginning of June 2005. So far SWM is ex-
instance. tremely satisfied with the results. That should
come as no surprise: “The improvements we
Step by step towards a mutual goal had aimed for have been greatly surpassed,”
summarizes IT expert Bayerl.
As this involved a pilot project, more time
than is usually the case was calculated for
its implementation. As the Head of System “Tests performed prior to productive deploy-
Operations H. Bayerl explains, “The result ment revealed that the computer performance
was important for us. Everyone involved was with Oracle RAC in Windows was higher than
basically sitting in the same boat and striving expected. While the dialog response times with-
towards the same objective.” out Oracle RAC were on average 450 milli-
seconds, these improved significantly with
All project steps were worked out according to Oracle RAC to reach around 340 milliseconds,”
a project plan that was drawn up beforehand. said system specialist Fischer.
Problem cases and platform-specific adaptations “This situation has led us to examine integra-
were approached jointly and solved with the tion of the SAP-HR system operating separa-
objective in mind. The first phase involved tely in the RAC environment, which would
setting up the new SAP-Oracle-RAC-R/3 signify savings in the administration costs.”
landscape including new hardware. The work
was carried out with current SAP data right But that is not everything. The availability
from the start. The second phase comprised has increased demonstrably, to be precise from
testing, while the system and individual 98.9 to 99.5 percent (within a monitoring
elements underwent continuous tuning. period of two months), which has important
“Specialists from the Development department consequences for SWM in view of its aim to
of Oracle USA assisted in work on the Oracle achieve as great a SAP availability as possible.
The BladeFrame allows an organization to
8

It goes without saving that the “fail-over func- Further Oracle use is definitely in planning
tionality” was extensively tested before the Thanks to the positive results experienced with
productive start. Three instances serve one Oracle RAC both for SAP use in Microsoft
database in parallel. A Windows cluster is Windows and when using non-SAP applications
operated on the three nodes in order to ensure on an open-source basis, the IT experts at SWM
that the central instance is also failsafe. aim to extend the use of Oracle Real Application
Clusters. That means there are plans for a
Furthermore, the new technology has made it migration of all smaller applications to Oracle
possible to increase scalability to a higher level, combined with use of Oracle RAC. In other
thus providing the benefits of extended invest- words, “we’ve shown that we can kill more or
ment safeguarding. This consequently enables less several birds with one stone using Oracle
an upgrade from a 3-node to a 4-node environ- RAC. The cost aspect plays a crucial role for
ment at any time. It is also possible to achieve this. Oracle RAC makes us hardware-indepen-
an additional system extension with compara- dent while enabling us to achieve the same or
tively inexpensive hardware. “The sky's the even more with less hardware.”
limit for us now,” aptly summarizes one mem-
ber of the IT department.

O R A C L E R E A L A P P L I C AT I O N C L U S T E R S ( R A C )
ADVANTAGES FOR SAP CUSTOMERS

Oracle RAC enables the Oracle Database Server to run real applications on clusters. By real
applications we mean that RAC does not only support applications designed with a clustered
database architecture in mind, but mainstream business applications of all kinds. This
includes popular packaged products such as SAP. It includes both OLTP (e.g. mySAP ERP)
and DSS (e.g. SAP BW).

High Availability and Workload Distribution the new server and the buffer cache in the stand-
by’s memory needs to be populated with the
One of the key drivers to move to Oracle RAC
most frequently used data.
is its compelling active/active architecture.
This architecture improves high availability and Oracle RAC’s active/active system offers virtually
provides better resource utilization. It reduces uninterrupted processing for users and elimina-
unplanned as well as planned downtime, but tes the need for these recovery steps – all cluster
even if no problem occurs, Oracle RAC allows nodes are always connected to all the storage so
customers to distribute the workload to as no volume dismounting and remounting is
many nodes as are available. required. Users on the surviving nodes remain
connected, only the users on the failed node
With a single SMP server, if anything fails the
need to reconnect.
whole machine goes down – user processing is
interrupted and can take 20 to 30 minutes to Oracle RAC helps to reduce planned down-
resume on a cold standby system (active/passive time as well. As RAC nodes can be added and
architecture). In such an environment, when the removed “on the fly,” maintenance tasks such
primary system fails, administrators have to as hardware or operating system upgrades can
manually dismount and remount storage volu- be performed on individual nodes without
mes, user connections have to be established to shutting down the system completely.
O r a c l e R e a l A p p l i c a t i o n C l uO
s trearcsl e( RDAaCt a
) bAadsvea 1
n0t agg–e sI nffoorr m
SAa tPi oCnu si st oCm
uer r es n c y 9

1a: Active/passive cluster solution (failover cluster) 1b: Active/active cluster solution (Oracle RAC)

Performance and Scalability Grid Computing (Adaptive Computing)

SAP’s applications ares based on a 3-tier archi- Today, businesses are putting IT organizations
tecture: the data are managed by a database under tremendous pressure to deliver the high-
server, the application functionality runs on est quality of service, in terms of maximum
an application server, and on the user’s device availability and scalability, at the lowest pos-
runs nothing but the presentation functiona- sible cost, and at the highest level of efficiency
lity. and flexibility. In short, IT is being asked to
do much, much more with much, much less.
This architecture provides for scalability at the
application layer, because SAP supports the Enterprise Grids make these seemly impossible
distribution of the application workload across challenges achievable. Enterprise Grids, made
several application server instances, which in up of large configurations of very low cost, com-
turn can be run on several application server modity clusters, dramatically reduce the cost
machines. of computer hardware. Oracle RAC technology
enables this low-cost hardware platform to deli-
However, SAP’s architecture is not as scalable
ver the highest quality of service that rivals and
at the database layer, as it assumes that there
exceeds the levels of availability and scalability
is one single database server only. Therefore, in
achieved by the most expensive, mainframe
the past, when the workload was growing, at
SMP computers. By dramatically reducing
the application layer it was up to you to either
administration costs and providing new levels
replace the existing machine by a bigger one
of administration flexibility, Oracle is enabling
(“scale up”) or add an additional machine of
the enterprise Grid environment.
similar size and power (“scale out”) – whereas
at the database layer the “scale up” approach The Enterprise Grid will have a profound im-
was your only option. Oracle RAC has chan- pact, enabling business to be more adaptive,
ged that: As for an application an Oracle RAC proactive and agile. With enterprise Grids,
system looks perfectly like one single database data centers will have the ability to meet the
server, the SAP application now accepts the changing demands of business at each moment
existence of several Oracle instances running in time – just in time. Application workloads
on several machines. In other words: Oracle will be managed as services that must meet
RAC gives you all the scalability options at defined levels of quality. Processing resources
the database layer which you have always had and storage will be allocated to services in a
on the SAP application layer. fluid fashion to ensure that those quality levels
are maintained. Each processing node or stor-
age component in the Grid can change its
individual personality almost instantaneously
without any downtime for any application.
10

“We have achieved all our objectives. Despite the fact that the solution is very
innovative, Fujitsu Siemens Computers implemented the project without any
hitches and faster than planned. That is not necessarily the rule in the IT sector.”

—RAINER WOHLHÖFNER
Head of Information Technology,
SRI Radio Systems GmbH

M A N A G I N G M A X I M U M AVA I L A B I L I T Y S I M P LY A N D E F F I C I E N T LY:

SRI RADIO SYSTEMS TAKES A NEW APPROACH TO


RUNNING SAP R/3

SRI Radio Systems


GmbH, Durach Powerful, reliable business processes are a top About SRI Radio Systems
priority for more and more companies. Yet this
Few mobile phone users know what they are,
usually means that costs grow faster than the
but use them just the same: wireless base
demands, for example due to redundant infra-
www.sri.de stations with UMTS and GSM technology from
structures and the time and effort needed for
The Customer
SRI Radio Systems (SRI). The company from
operations management. That is why SRI Radio
SRI Radio Systems GmbH, Durach, a suburb of Kempten, Germany,
Systems took a different approach, introducing
Durach employs just over 500 people and has long
a solution that comprises a unique combination
been an established partner of carriers. The
The Project of leading software on Solaris-based PRIME-
latter have come to respect the high standard
Optimization of the POWER servers and is distinguished by fault
IT infrastructure for SAP®
of quality of the systems, for example, and the
tolerance, performance on demand, optimal use
R/3® to ensure higher
refined production and logistics processes with
of resources and efficient management. For the
and simpler operations which SRI ensures its ability to deliver at
first time in the world, Fujitsu Siemens Compu-
management short notice. The fact that the company lives
ters implemented SAP R/3 on Oracle Real
up to the very highest standards was confir-
The Solution Application Clusters (RAC) with VERITAS
med when it was named “The Best Factory”
The world’s first Solaris- Storage Foundation™ for Oracle RAC (SFRAC)
based implementation
in the Industrial Excellence Award 2005.
from Symantec Corporation (formerly VERI-
of SAP® R/3® on Oracle TAS Software).
Real Application Clusters;
installation and manage-
ment with VERITAS Storage Benefits for SRI Radio Systems The solution at a glance
Foundation™ for
• Maximum availability for business processes • Introduction of a disaster-proof cluster environment for
Oracle RAC from
• No unused server capacities SAP R/3
Symantec Corporation
• Lower administration overhead • Migration from SAP R/3 4.6c to Oracle RAC
Partners:ey Benefits: • Simple scalability of the environment • Implementation and management with VERITAS Storage
Oracle and Symantec • Improved response times Foundation™ for Oracle RAC
Cost Reduction • Rapid implementation of the new solution • Servers: 3x PRIMEPOWER 450, each with four SPARC™
Long-term IT concept in less than three months processors and 16 GB main memory
to ensure ROI • Operating system: Solaris
Streamlined DBA
• Concept: SRI Radio Systems and Fujitsu Siemens Computers
• Implementation partners: Oracle and Symantec Corporation
SAP on Real Application Clusters (RAC) at S.R.I. with Fujitsu Siemens 11

Optimization according to the minimax Here are the details of the solution, which has
principle been tested by SAP exclusively for Fujitsu
Siemens Computers. One Oracle instance and
Whether in Rovaniemi, Durban or Hamilton, an application server are installed on each
SRI supplies and installs turnkey base stations system in the cluster. The enqueue server and
worldwide within just five days on request – a the replicated enqueue have also been separated
service that would be unimaginable without and virtualized. SRI uses VERITAS SFRAC
perfect production and logistics processes. for cluster management and switchover. The
And the availability demands that SRI makes impact of this combination is enormous:
of its entire SAP environment are accordingly maximum availability, yet minimum admini-
high. However, they were hard to reconcile in strative overhead and costs. There is reliable
practice because the production line runs five protection against individual potential error
or six days a week over three shifts. That left sources. The entire environment is also disaster-
only a very narrow time frame on the wee- proof, since one node is installed at a separate
kend for maintenance work. “Any location. Optimal use is made of computing
problems there would have had an immediate power. SRI was also able to reduce its hardware
impact on our business,” says Rainer Wohlhöf- investments, since no redundant systems are
ner, Head of Information Technology at SRI. required. This would not have been possible
That was a risk that SRI wanted to minimize with conventional clusters. And, last but not
with a new solution offering maximum availa- least, users profit from response times that are
bility for the SAP environment and simpler better by around 40 percent. So it is no wonder
operations management. that Wohlhöfner is satisfied with the new
solution: “The concept is very convincing. We
Fujitsu Siemens Computers offered the ideal also liked the fact that Fujitsu Siemens Com-
concept with a high-availability parallel com- puters needed less than three months for the
puting environment. project and that the changeover was complete-
VERITAS SFRAC and SAP R/3 on Oracle ly transparent. Our 480 employees didn’t
RAC were implemented on three Solaris-based notice a thing.”
PRIMEPOWER servers. And there were good
reasons for this choice. With its multiple,
simultaneously active nodes, Oracle RAC
ensures maximum fault tolerance for the data-
base. Since the workload is distributed across
the cluster, companies can use several smaller
and low-cost servers and, if needed, expand
the environment flexibly by adding new
servers. Effective support is offered here by
VERITAS SFRAC. Components of the inte-
grated solution include special cluster server
technology and a cluster file system to enable
simpler installation and management of RAC
databases. In the event of an error, this ensu-
res that there is a smooth switchover between
the servers; one of the systems remains the
master and data consistency is retained.
12

“The SAP infrastructure changeover has completely fulfilled our expectations. We can benefit from the
SAP-validated and certified infrastructure architecture model with Oracle RAC and Linux in a variety
of respects for SAP NetWeaver BI-/SAP-SEM use.”

— JOSÉ LUIS CHECA,


Systems & Architecture Manager,
Gas Natural

S A P N E T W E AV E R B U S I N E S S I N T E L L I G E N C E / S A P
S T R AT E G I C E N T E R P R I S E M A N A G E M E N T
AT A G L A N C E
Gas Natural has utilized a system platform chan- And on top of that, we also wanted to be able to
geover to optimize SAP NetWeaver BI-/SAP SEM respond to a more intensive future use of the SAP
use and create advantages in terms of efficiency solution in a more flexible and straightforward way.”
With approximately 10 million customers, the Gas A significant trigger for optimizing the system per-
Natural Group based in Barcelona represents a key formance was the batch and online operation occurring
international player in the energy sector. The Gas at the same time, in which extremely large data
Natural Group, operating in a total of 10 countries volumes had to be processed in a very short time. This
with just under 6,700 employees, achieved annual mixed operation was called for in particular because
net profit 633.9 million euros in the financial year of increased processing requirements with the SAP
2004. Since deregulation of the energy market in SEM consolidation of budget/finance data in Latin
1998, the multinational company has been pursuing America. The total worldwide SAP environment of
an ambitious expansion and diversification strategy. Gas Natural with over 5000 users is served centrally
In addition to the production, supply and transport from Barcelona, 7 days a week X 24 hours a day from
of natural gas, the energy group now generates elec- 2 major computing centers.
tricity with its own power stations and wind turbine
generators, which it provides to end consumers and Added value through a combination of the
companies. To increase competitiveness and realize
tried-and-tested with the new.
growth-oriented company objectives, Gas Natural has
intensively utilized solutions provided by the mySAP A taskforce consisting of members of the Gas Natural
Business Suite family since 2000. IT section examined solution alternatives for achieving
“SAP solutions are an important cornerstone for us on the objectives set out and evaluated these in detail.
the path towards further profitable growth,” states The following innovations were adopted, which were
José Luis Checa, Systems & Architecture Manager at to be implemented as part of a project:
Gas Natural. • Implementation of a horizontal architecture model
for extending the database level based on Oracle
Several objectives on the to-do list
Real Application Clusters 9i with simultaneous
In the autumn of 2004, Gas Natural introduced an improvements on the application level, without
extensive realignment for the system platform/archi- changing these technically.
tecture utilized, with the specific aim of optimizing
• Validation and certification of the infrastructure
use of the combined application solutions SAP
architecture by SAP.
NetWeaver BI-/SAP-SEM. As IT Manager José Luis
Checa explains, “To begin with, there was a real need • Introduction of the uniform operating system
for action to get the system performance and response platform Linux Red Hat with newIntel-based
times up to a better level. At the same time, we also hardware on the application and database level for
endeavored to improve the cost efficiency for SAP NW SAP NW BI-/SAP SEM use; but without funda-
BI-/SAP SEM operation. mentally modifying the SAP solutions.
SAP on RAC at Gas Natural 13

• Migration of the existing SAP NW BI-/SAP SEM Impressive productivity and efficiency boosts
database to the new RAC database cluster Linux Mr. Checa sums up what has been achieved as follows:
platform. “The SAP infrastructure changeover has completely
• Extensive tests, so as to be able to check functional fulfilled our expectations. We can benefit from the
correctness and carry out performance measure- SAP-validated and certified infrastructure architecture
ments for the old and new environment. model based on Oracle RAC and Linux in a variety
of respects for SAP NetWeaver BI-/SAP-SEM use.”
According to IT expert Mr. Checa “we opted for RAC Measurements carried out by Gas Natural have thus
as we had positive experience with the Corporate Data revealed that “the SAP NW BI/-SAP SEM response
Warehouse used since 2000 and which currently has a time have improved by an average of 144.5 percent
volume of 2.2 TB. At the same time, we were also thanks to the use of database clusters. The level of
striving towards improving system performance. satisfaction among users has increased considerably.”
The decision for Linux was justified on the basis that Furthermore, the energy group can now realize
a more cost-effective platform was associated with the potential platform extensions, which may be neces-
SAP NetWeaver BI-/SAP SEM in comparison to sary because of increasing SAP NW BI/-SAP SEM
the Unix system used up until now – in particular user numbers. “All we do then is bring further RAC
through the use of standard Intel/-AMD servers.” nodes or servers into operation,” states Mr. Checa.
In the Corporate Data Warehouse strategy of Gas
Short project realization time – great results from Natural, SAP NW BI/-SAP SEM (currently over 700
the partners involved named users) represents a hugely significant element,
The optimization project was realized in a time frame which is becoming increasingly more important for
of just 3 months – including validation and certifi- the entire company operation. In addition, the SAP
cation of the infrastructure model. The new SAP NW database availability has improved further because of
BI/-SAP SEM system platform has been running since the new system platform. If one (or more) RAC nodes
January 2005. The system utilizes a 2-node Red-Hat fail for some reason, the others immediately assume
cluster with two Intel-based HP servers on the appli- their task(s). The three database administrators
cation level. responsible for the databases at Gas Natural are also
The database level consists of a 7-node cluster with responsible for the RAC implementation.
Linux Red Hat 3.0 and Oracle RAC 9.2.0.5 as well
as the Oracle Cluster File System 1.0.13 on 7 Intel- Cost reduction through Linux
based HP servers. Formerly, Windows (Intel server) When it comes to the Linux platform, Gas Natural is
was used on the application level and Unix (Unix equally satisfied with the results achieved. More pre-
high-performance computers without RAC) on the cisely: “Linux allows us to significantly reduce the SAP
database level. In addition to the SAP validation/cer- operating costs. Instead of special Unix servers, we can
tification and the extension of the new SAP system now utilize more cost-effective standard Intel/AMD
platform, the project work focused on database migra- servers. In comparison to Unix, the cost ratio is at
tion (size of the database: around 290 GB) towards least 1:10,“ stresses IT expert José Luis Checa. Gas
RAC as well as extensive tests with special test tools. Natural also wants to extend RAC/Linux to the SAP
As Gas Natural IT Manager Mr. Checa recalls, IS/U application in future, in addition to the SAP
“Everyone involved pulled together wonderfully for NetWeaver Business Intelligence/SAP Strategic Enter-
the work. That meant great results. The effective prise Management. The objectives here: to reduce the
teamwork between SAP, Oracle, HP and Red Hat time frame with batch runs for accounts settlement;
was a crucial factor for the success of the project.” but at the same time have recourse to cost-effective
standard servers, or improve system scalability against
“The use of database clusters has meant that SAP the backdrop of continually increasing numbers of
NW BI/-SAP SEM response times have improved customers. The planning for this is already underway
by an average of 144.5 percent.”
in the multinational and expansion-oriented energy
group.
14

Summary • Extension of the new SAP-RAC-Linux environ-


ment together with new Intel server hardware
Gas Natural has optimized SAP NetWeaver BI-SAP
and changes on the database and application level.
SEM use on the basis of Oracle Real Application
Clusters (RAC) and Linux. Thus, the multinational • Close coordination between SAP, Oracle, HP and
energy group has boosted system performance, Red Hat.
increased IT cost efficiency and can also respond to • Extensive tests for the new SAP-RAC-Linux
further SAP NW BI-SAP SEM use very flexibly in platform.
terms of its IT capabilities.
Main benefits for the customer:
Sector • Above-average improvement in response times;
Energy service sector the level of satisfaction among SAP NW BI-/SAP
SEM users has increased demonstrably.
Internet address: • Future security through SAP validation and
www.gasnatural.com certification of the new architecture model.
• Simplified scalability for SAP solutions based on
The most important challenges: databases; more flexible SAP NW BI-/SAP SEM
• Improvement in the SAP NetWeaver BISAP SEM extension.
system operation resulting from a change to the • Cost savings with the SAP infrastructure platform.
user application, in particular via an overlapping
• Basis for the use of cost-effective Intel/AMD-
of batch and online processing
based “standard servers.”
• Cost reduction in the SAP infrastructure.
• No general changes required for the SAP software.
Project aims: • Further increase in the SAP high availability.
• Introduction of new infrastructure technology
Existing System landscape:
components for optimized SAP NetWeaver (NW)
BI-/SAP SEM use, with significant system or per- • SAP R/3 Enterprise (4.7) and SAP IS-U (4.64)
formance improvements. • SAP NetWeaver Business Intelligence 3.1 and
• Strategic realignment of the SAP infrastructure SAP Strategic Enterprise Management 3.2
for a more cost-efficient and future-flexible IT
expansion, based on a SAP-certified infrastructure Database
architecture model and standard IT Components, • Oracle Real Application Clusters (RAC) 9.2.0.5
without farreaching changes on the application with Oracle Cluster File System (OCFS) 1.0.13
level.
Hardware
Decision for Oracle RAC and Linux: • Hewlett-Packard HP
• Experience already gained with the Oracle Real
Applications Cluster (RAC) solution. Operating system
• Realization of cost savings via a platform • Linux Red Hat AS 3.0
migration towards open source/Linux.

Highlights of the implementation:


• Formulation of feasibility studies with concrete
project stages.
• SAP certification for the new infrastructure
architecture.
Certified SAP on RAC Configurations for Windows and Linux 15

C E R T I F I E D S A P O N R A C C O N F I G U R AT I O N S F O R W I N D O W S A N D R A C
We have separate scenarios for Windows and Linux as well as for RAC 9.2 and RAC 10.2 (in controlled availability from 04/2006
on i.e. for selected customers)

Windows Linux

1. Oracle RAC 9.2 on Windows x86 and IA64 (no support for x86_64) 3. RAC 9.2 on Linux x86, x86_64, and IA64 for RHEL4 or SLES9
• Oracle 9.2.0.7 or 9.2.0.8 • Oracle 9.2.0.7 or 9.2.0.8
• OCFS 9.2.0.7 or 9.2.0.8 for database, Oracle software and SAP software • OCFS2 1.2.0 or higher for database, Oracle software and SAP software
• Windows 2003 Release 1 (no support for Windows 2003 Release 2) • NetAppliance NFS for database, Oracle software and SAP software
minimum 2 nodes
• RHEL 4.2 or higher
• SAP Enqueue Replication covered by Microsoft Cluster Services
(SAP Central Instance can be co-located with RAC instances) • SLES9 SP3 or higher

• iSCSI only supported up to 4 nodes • minimum 2 nodes without SAP Enqueue Replication
• minimum 4 nodes with SAP Enqueue Replication (2 nodes for RAC
Not supported: and 2 seperate nodes for Red Hat Cluster or HP ServiceGuard Cluster
– crossover cables for interconnect (only Gigabit Ethernet, minimum for SAP Enqueue Replication)
two cards) • SAP Enqueue Replication covered by HP ServiceGuard or Red Hat
– Infiniband Cluster on two separate nodes (SAP Central Instance cannot be
– raw devices co-located with RAC instances on same node)
2. Oracle RAC 10.2 on Windows x86, x86_64 and IA64 (from 04/2006 Not supported:
on for selected customers) – iSCSI – Firewire – crossover cables for interconnect (only Gigabit
• Oracle 10.2.0.2 or higher Ethernet, minimum two cards) – Infiniband – Red Hat GFS
– Veritas CFS – ASM – raw devices
• OCFS 10.2.0.2 or higher for database, Oracle software and SAP software
• Windows 2003 Release 1 (no support for Windows 2003 Release 2) 4. Oracle RAC 10.2 on Linux on x86, x86_64 and IA64 for RHEL4 or
minimum 2 nodes SLES9 (from 04/2006 on for selected customers)
• SAP Enqueue Replication covered by Microsoft Cluster Services • Oracle 10.2.0.2 or higher
(SAP Central Instance can be co-located with RAC instance)
• OCFS2 1.2.0 or higher for database, Oracle software and SAP software
• SAP Enqueue Replication covered by Oracle Clusterware (SAP
Central Instance can be co-located with RAC instance on same node) • NetAppliance NFS for database, Oracle software and SAP software

• iSCSI only supported up to 4 nodes • RHEL 4.2 or higher


• SLES9 SP3 or higher, minimum 2 nodes without HP ServiceGuard
Not supported: or Red Hat Cluster (Oracle Clusterware for SAP Replicated Enqueue),
– crossover cables for interconnect (only Gigabit Ethernet, minimum minimum 4 nodes with HP ServiceGuard or Red Hat Cluster for SAP
two cards) Enqueue Replication (2 nodes for RAC and 2 seperate nodes for Red
– Infiniband Hat Cluster or HP ServiceGuard Cluster for SAP Enqueue Replication)
– ASM • SAP Enqueue Replication covered by Oracle Clusterware (2 node con-
– raw decives figuration, co-location of SAP Central Instance with RAC Instance)
• SAP Enqueue Replication covered by HP ServiceGuard or Red Hat
Cluster (minimum 4 nodes are required, SAP central instance cannot
be co-located with RAC instance on same node)
Not supported:
– iSCSI – Firewire – crossover cables for interconnect (only Gigabit
Ethernet, minimum two cards) – Infiniband – Red Hat GFS
– Veritas CFS – raw devices
16

With this slogan, Dr. August Oetker Nahrungsmittel KG produces and markets a wide variety of products
under the brand name established since 1891. In developing the first storable and ready-made baking
powder in 1893, Dr. August Oetker laid the foundations for a corporate group which currently operates
on a successful international basis with around 21,000 staff in six different business segments.

“QUALITY IS THE BEST RECIPE”

With Dr. August Oetker KG as a holding company, For Mucha and his team, this meant an unjustifiable
the Oetker Group is also active in the beer and spark- extra workload as a result of programming and help
ling wine area as well as in shipping and financial programs. Reliable reporting was also only possible
services In addition to the food business segment. in certain circumstances or by indirect means in this
Companies from the chemicals industry, publishing scenario. A reorganization of the system architecture
and the five luxury hotels of the Oetker Group are appeared unavoidable.
brought together in a further business field.
Migration to new territory
The Oetker Group achieved a turnover of 6.4 billion
euros in the financial year 2004 and is poised for In the summer of 2003, Oracle was given the chance
further international growth. to prove that the problems extant could be solved
through a migration of the existing SAP BW to
A business warehouse with minor faults Oracle9i.
The test migration from DB2 7.2 to Oracle9i turned
Preparations have been underway for the final aban-
out to be more difficult than initially assumed. This
donment of the mainframe world since 2001 at
reason for this lay in the fact that, in contrast to SAP
Dr. Oetker. The necessity of conversion from OS 390
R/3, SAP BW had been developed using a database-
and DB2-based computer to Unix was out of the
dependent approach in certain areas while the SAP
question here.
migration tools (R3load) did not take these depen-
In 2003 it became clear that an anticipated and inter- dencies sufficiently into account. Despite these chal-
nationally available business warehouse (BW) solution lenges, the results have convinced Dr. Oetker to move
with DB2 UDB based on SAP was rapidly reaching to the database platform.
its limits. The enormous growth in the database led
In December 2003, the Oetker Group concluded a
to bottlenecks in the system extension. Tables were
license agreement with Oracle for 3000 named users,
sometimes defective and the system activation was
in order to replace the existing 53 DB2 installations.
inconsistent in its execution. A typical result of this
The project began with the above SAP BW, the pro-
was that no attributes could be added to strategically
ductive conversion of this to Oracle 9.2 occurring in
important tables for recording and updating sales
the first six months of 2004.
customer data. Deadlocks were the order of the day.
After the success of these first migrations, the
“We had to allow for at least one man day per week
systems of the Radeberg brewery and of Henkell &
on average for solving everyday problems, both in
Söhnlein will migrate to Oracle in this and the
regard to the database and for the applications, which,
forthcoming years in addition to the ERP systems of
of course, was not a sustainable situation,” states
Dr. Oetker Nahrungsmittel KG.
Jörg Mucha, IT Organizer in the areas of Organiza-
tion, System Development and Business Intelligence. The advantages of a SAP business warehouse whose
operation is based on Oracle gradually became all the
more visible. Despite the very complex systems with
over 400 Infocubes, the problems which constantly
occurred when using the previous database, such as
optimizing and locking, vanished into thin air.
D r. O e t k e r, D B 2 t o O r a c l e m i g r a t i o n f o r S A P B W 17

At the same time, the Oracle database proved to be As the manager responsible, Frank Pickert, IT Head
more economical in disk usage. Another concept in of Dr. August Oetker Nahrungsmittel KG, is now
saving numerical values and the use of bitmap indi- convinced of the merits of Oracle: “The growing
ces, which the SAP BW only uses with Oracle as complexity of the applications at Dr. Oetker calls for
standard, enabled disk space requirements to be an operationally reliable database system with corre-
reduced by around 20 percent. The path towards sponding growth potential. We regard the Oracle DB
achieving this proved to be “stony” in many res- as a strategic platform for integrating our affiliated
pects, especially as the migration from DB2 to companies in a uniform BW, ERP and CRM frame-
Oracle and SAP BW represented a voyage into work as well as for all further SAP and non-SAP
uncharted territory. applications extending beyond these modules.”
A team of specialists from the Oracle Support and It was and is a good tradition at Dr. Oetker to always
Solution Center in Walldorf, SAP experts and the opt for the products and services of the relevant
customer Oetker succeeded in realizing the migra- market leaders. This principle has been confirmed
tion within a time period which was very reasonable yet again with the migration from DB2 to Oracle9i,
given the complexity of such an application. At the as Oracle remains the number one choice worldwide
same time, SAP had to supplement the existing for 70 percent of all SAP applications.
tools with newly developed reports which, after It will come as no surprise then that all CHESS
transfer of the data from DB2 to Oracle, converted systems are based on the Oracle database.
the database in such a way as though the SAP BW
system were developed and setup with Oracle right
from the start. The process developed at Dr. Oetker
has meanwhile been successfully implemented with
further customers who have realized a similar migra-
tion. After the introductory phase, the system now
works with full stability at Dr. Oetker and is set to
be extended in the near future.
“We are very satisfied when it comes to operational
reliability” is how Dr. Franz Brüggemann, IT Busi-
ness Manager at Dr. Oetker, describes the current
situation. “We've finally eliminated the deadlock
problems. This and many other aspects were crucial
in our decision to incorporate Oracle as an integral
component in our entire CHESS business warehouse
in future”.

Harmony in an international environment


In the next few years, the company is endeavoring
to harmonize all systems in the international group.
A new UNICODE-based environment with R/3, a
BW and a CRM-System, in which the previous
systems are consolidated, were created under the
project name CHESS (Consolidation and Harmoni-
zation of the European SAP Systems). CHESS has
been productive since early 2004 and is continually
expanding through the integration of individual
national systems. The experiences gained from opera-
tion of the existing SAP BW systems was taken into
account when planning CHESS.
18

Amazing Improvements

Performance problems in response to constant increase in SAP usage and the announce-

ment that SAP would not support Unicode functionality for Informix forced the company

Testo to take action in its database utilization. A successful Informix-Oracle migration has put

this producer of professional test equipment in good hands. Various advantages were real-

ized by the platform change and not just those measured in monetary value.

T E S T O A G I N G E R M A N Y M I G R AT E S T H E I R
D ATA B A S E F R O M I N F O R M I X T O O R A C L E F O R
THEIR SAP SYSTEM

From the time Testo was founded in the year Nevertheless, where the operating system and
1957 until today the company has conti- database were concerned Testo continued with
nuously grown from a small to medium size what was tried and true. “We had built up
business and currently employs about 1,300 extensive Tru64 and Informix know-how that
people worldwide (2004 sales: 105 million we wanted to continue to be able to exploit
euros). In recent years the expansion-oriented that” explained Alexander Bühler of the Testo
Testo AG
Lenzkirch, Germany
company has achieved especially large gains in IT team. For some time now Testo has employ-
growth. It was not without reason that the ed various tools of the mySAP Business Suite:
Industry: “Growth Plus Testo” corporate group was ran- SAP R/3 (4.6C) with the components FI, CO,
Specialist for ked among the “500 mid-cap companies with MM, PP, SD, QS and HR as well as mySAP
Measurement Technology, the strongest growth” last year. Today the CRM (4.0) and SAP BW (3.2). Applications
23 Subsidiaries company from the Southern Black Forest has for e-mail, fax and telephone are also used.
operations worldwide with 23 subsidiaries. The HP hardware with Cluster File System
Annual Revenue: consists of two ES40 units, each with 4 Alpha
2004, 105 M€ Testo has always used its IT to efficiently CPUs and 10 GB memory.
shore up its growth while also using it to
Employees: continuously improve efficiency at the com-
1,300 pany. For about 10 years the test equipment
specialist utilized an ERP system from the
Oracle Product: Bäurer company together with HP hardware
Oracle Database based on the Tru64 operating system and a
9.2.05 database from Informix. In early 2002 Testo
Key Benefits: turned away from the Bäurer ERP software
Improvemen and switched to SAP R/3, primarily to safe-
guard its future needs especially in considera-
tion of its new international orientation.
I n f o r m i x - O r a c l e d a t a b a s e m i g r a t i o n a t Te s t o 19

Negative aspects forced company to act Speaking in favor of Oracle according to Testo
was the fact that the prepared database is
As he reports: “The first problems surfaced
available on most operating systems, and as a
after about a year and a half with regard to
stand-alone database it is not preloaded on the
system stability and performance. We had
hardware side. The analysis also indicated that
interfaced a number of subsidiaries to our
a migration to Oracle would not require the
central SAP system, and this led to rapid
purchase of any newer, faster hardware. Further-
growth in data. Users complained again and
more, the Oracle RAC (Real Application
again about long, unacceptable response
Clusters) would offer the option of additional
times. The reason for these difficulties was – Alexander Bühler,
scalability, also without requiring additional
our Informix database.” Of course Testo could
hardware. Another argument Testo offered in SAP Basis Administrator
have gotten a handle on the problem to some
favor of Oracle was said that SAP itself uses Testo AG
extent by means of additional server equipment,
the database for productive systems, and good
but the company did not want to make any
and meaningful feedback was obtained from
cost-intensive investments in hardware.
various customers who were given as referen-
When another disadvantage of the database
ces. That is how Bühler of the Testo IT team “ The upshot of this heightened
materialized the company felt compelled level of satisfaction and the
portrayed the decision-making process.
to act. Testo wanted to utilize the Unicode
capability offered by SAP, because it was corollary reduction in demand
expanding with large subsidiaries in Asia and Smooth database Migration – without
for our support is that the IT
this required multi-byte representation of any stumbling blocks
team, especially the admini-
character sets. “After the IBM takeover of The Informix-Oracle migration ran like
Informix, SAP announced that it would no clockwork and according to plan. strators, can now devote their
longer support Unicode capability for Infor- The key work steps were executed one after energies to other important
mix” explained Bühler. That sealed long-time another: First the creation of a client copy of
Informix user Testo’s decision to migrate to SAP R/3 and a client copy of mySAP CRM,
tasks, projects for example ”
a different SAP database system. then transfers to a test system with Oracle,
a test migration with SAP Tools (Run time:
The company scheduled three months for an About 50 hours), and complete integration
evaluation or market analysis for the purpose testing (including subsidiaries).
of selecting the database system Testo would
purchase for its future SAP operations. According to Testo the migration “…was
According to comments by SAP expert Bühler tested for about three weeks.” During that
“The Microsoft Windows operating system time period we were only able to identify
was excluded from the outset, and therefore three faults, and it was possible to correct
also the Microsoft SQL Server database. And these rat her quickly.”
IBM’s DB 2 was not under serious considera-
tion either. Therefore we focused our evalua-
tions on the Oracle and MaxDB databases.”
20

monthly overview October monthly overview November

processor load in the course of the day 20/10/2004 processor load in the course of the day 18/11/2004
I n f o r m i x - O r a c l e d a t a b a s e m i g r a t i o n a t Te s t o 21

Afterwards work began on the production Yet another improvement realized by database
migration that was to be executed over a migration, and one that is very meaningful
weekend. The migration was started on a for the mid-size company, is this: The IT
Friday evening about 7:00 pm, and it was team hardly gets any more phone calls from
completed by Sunday evening at 6:00 pm. users with regard to performance problems.
The primary tasks were: SAP Export from “The upshot of this heightened level of satis-
production system using SAP Tool R/3exp faction and the corollary reduction in demand
(18 hours); in parallel, installation and confi- for our support is that the IT team, especially
guration of Oracle on the production system; the administrators, can now devote their ener-
SAP Import to the production system using gies to other important tasks, projects for
SAP Tool R/3load (6 hours); startup of the example”, stated Bühler.
production systems and testing. No errors of
any kind were found, and the startup of the In his concluding remarks he revealed that
SAP production systems ran smoothly and it is not just the SAP users who are satisfied
without any problems. with the Informix-Oracle migration; manage-
ment too is satisfied in light of the financial
Funds for further IT implementations advantages based on the fact that Oracle made
hardware expenditures obsolete. This meant
“In contrast to our Informix experience the that funds could be invested in new SAP pro-
differences were quite astounding” was the jects, e.g. in the introduction of mySAP ERP
assessment of SAP expert Bühler. He added: for which Testo already has firm plans.
“Oracle is currently saving us a lot of money.”
That is because in spite of further increases in
data volume – and with the same hardware
and same CPU – usage loading of the systems
has quite simply been reduced. “Before Oracle
came along the load on the database server
was between 60 and 70 percent. Today it is
between 20 and 40 percent. And the load on
the application server has also been reduced,”
according to analyses by Testo. Before the
Informix-Oracle migration the data volume
was approx. 290 GB, whereas currently is at
330 GB.

Testo has an appreciation for the significance


of this reduction in server load. The bottom
line is that the company will not need to per-
form any cost-intensive hardware upgrades
in the foreseeable near future. And if it does,
it will always be possible to use Oracle-RAC
technology which distributes load to multiple
computing nodes.
22

O R A C L E 9 i D ATA M I N I N G C O N N E C T O R 1 . 1
F O R m y S A P TM B W

The ODM Connector enables users of SAP ® Business Warehouse 3.5 to mine data seam-
lessly using Oracle9i Data Mining within the SAP Data Mining framework. This SAP-certified
solution gives users the powerful and scalable Oracle9i Data Mining in-database algorithms
to extend and complement the SAP analytics offering.

ODM CONNECTOR 1.1 Mining in the database provides SAP customers


With the seamless integration of Oracle9i with increased data security, as data need not leave
Data Mining and SAP BW 3.5 and SAP CRM, the database. Moreover, in-database mining is
the ODM Connector provides SAP customers scalable using Oracle’s Real Application Clusters
advanced analytics embedded in the Oracle9i (RAC), allowing all the data to be mined, not
database. Data and business analysts benefit just a sample. Avoiding copies of data, SAP
from the wide range of state-of-the-art data customers can maintain a single version of truth.
mining functionality that is exposed through
the SAP Data Mining framework. The ODM Connector allows customers to
leverage their investment in Oracle as most
With BW 3.5, SAP released an API that third SAP customers use the Oracle Database.
party vendors can use to connect their solution Because Oracle9i Data Mining is embedded in
to the SAP Data Mining framework. Oracle has the Oracle9i Database, mining results can also
certified Oracle9i Data Mining with the official be accessed through other Oracle tools, inclu-
BW-DMI 3.5 interface of SAP AG through the ding Discoverer, OLAP, and Reports. Using
ODM Connector. Integration saves SAP users Oracle9i Data Mining, SAP customers can
time and money to leverage Oracle’s in-database achieve an overall lower total cost data mining
mining capabilities. solution for analysis, development, and deploy-
ment while being fully integrated with their
The ODM Connector integrates all of the SAP environment.
Oracle9i Data Mining algorithms and provides
model viewers using the Oracle9i Data Mining SAP BW 3.5
Browser. The Connector automatically maps With BW, SAP delivers ready-to-go and easily
SAP data types and transparently performs extensible reports, data infrastructure, and
clean up and recovery actions. mining models for selected best business
practice scenarios. The immediate benefits
of using this business content are reduced
implementation time and costs.

BW provides the Analysis Process Designer,


which enables a workflow-based, graphical
definition of the data mining activity. Users
choose from a palette of tools for identifying
data to mine, transformations on that data,
and mining activities, e.g., model building.
Oracle Data Mining algorithms are accessed
Oracle Data Mining for SAP customers 23

SAP BW Analysis Process Designer – workflow-based SAP BW Data Mining Workbench


specification of transformations and mining

All Oracle9i Data Mining algorithms are made


The model viewing components of the Oracle9i
available in the ODM Connector as highlighted
Data Miner Browser are also integrated with
in the Instance Upgrade.
the ODM-BW Connector. This enables users
to see the model details as defined by Oracle9i
SAP BW Instance Upgrade
Data Mining. For example, in the figure below
we see the rules produced by the Adaptive
Bayes Network algorithm.

SAP BW Data Mining Browser – model viewer

As models are defined and built, they appear


in the Data Mining Workbench. Here the list
of available models and their status along with
the identifying vendor name.
24

T E C H N I C A L S P E C I F I C AT I O N S • ABN supports three modes of operation: build


a pruned Naïve Bayes model, a single-tree
Data Preparation model, or a boosted multi-tree model.
• BW-provided transformations • In the single-tree model, ABN provides
• Automatic data binning as part of model model transparency with human readable
build and apply (scoring). rules, e.g.:
IF MARITAL_STATUS = “Married“
In-Database Mining AND EDUCATION_NUM = “13-16“
Platform Requirements • All model build and apply operations THEN CHURN= “TRUE“
Oracle9i R2 Data Mining occur within the Oracle9i Database.
runs in Oracle9i R2 Data- • Default settings for all Oracle9i data mining Model Evaluation
base on all supported
operations. Ability to override and specify • Confusion matrix for Naive Bayes and
platforms.
settings. Adaptive Bayes Networks models
SAP BW 3.5 SP05 • Lift calculation for Naive Bayes and
including SAP J2EE Supervised Learning Adaptive Bayes Networks models
Application Server 6.40 • Prediction of binary or multi-class outcomes
is required. with supporting probability (e.g., assuming Unsupervised Learning
A, B, C, or D as possible outcomes, B will
occur with 60% probability). Association Rules Algorithm
• Real-time scoring performance for individual • Find the occurrence and likelihood of co-
records. occurring events, e.g., Q, R, and S are
associated with Z, 452 times, with 78%
Attribute Importance confidence.
• Identify attributes as important for predicting • Finds rules with support greater than a
a target attribute using Minimum minimum support and confidence greater
Description Length algorithm. than a minimum confidence.
• Most applicable for explicit attribute selection • Filtered rules retrieval using ODM Browser
for Naïve Bayes. Automatic attribute
importance/selection preprocessing is in- Clustering Algorithms

cluded in the Adaptive Bayes Network • Enhanced k-Means Algorithm: The hierarchi-
algorithm. cal, distance-based k-means algorithm
partitions the data into a predetermined
Naive Bayes Algorithm number of clusters. The k-means algorithm
• Makes predictions using Bayes’ Theorem, supports only numerical attributes.
which derives the probability of a prediction • Orthogonal Clustering (O-Cluster)
from the underlying evidence, as observed Algorithm: Creates a hierarchical, grid-based
in the data. clustering model. The resulting clusters
• Naive Bayes cross-validation permits the define dense areas in the attribute space.
user to test model accuracy on the same Clusters are described by intervals along
data that was used to build the model, rather the attribute axes and the corresponding
than building the model on one portion of centroids and histograms. O-Cluster supports
the data and testing it on a different portion. both numerical and categorical attributes.
It is particularly good in high dimensional
Adaptive Bayes Networks Algorithm space (many attributes).
• Like Naive Bayes, Adaptive Bayes Network • Example results: Records 12, 15, and 25 are
(ABN) builds models based on counts members of Cluster_1. Output includes:
observed from the data. cluster centroids, histograms per attribute,
and cluster rules.
Delta Consulting: RAC Services for SAP customers 25

D E LTA C O N S U LT I N G

Delta Consulting, Inc. is an SAP-focused consultancy offering extensive SAP consulting


services and Software Components for it’s SAP Customer Base. Delta has a strong
commitment to providing innovative and practical solutions that combine real-world
business expertise and products designed to extend and enhance the value and performance
of a customer’s SAP investment.

Our consulting services utilizes an extensive Services to be provided by Delta Consulting


database of senior consultants which average include:
over 9 years of SAP specific experience that
spans all SAP solutions and business services • RAC Implementations
that address the functional, technical and • Database Migrations Delta
infrastructure requirements of an SAP solution. • Tuning and Performance Optimization For additional
In additional, through our extensive Partner • System Healthchecks information regarding
network we can provide worldwide deployment • Backup/Recovery and Hot Standby Solutions Delta Consulting,
capabilities in the development area. please visit our website
Over two-third’s of SAP’s customers run their at www.go-delta.com
As a certified SAP Services Partner for over applications on Oracle. With SAP’s endorse- or contact
Jack Tomb,
10 years and a member of the Oracle Preferred ment and certification of Oracle9i RAC the
Partner at
Partner Program (OPP), Delta Consulting is motivation to implement RAC includes: 610-558-1730.
pleased to announce its selection by Oracle
to extend its agreement to provide consulting • Higher Availability
services for Oracle Real Application Clusters – New Requirements
(RAC). Using SAP with Oracle RAC offers a – Maintenance window negotiation
unique solution for SAP customers already – More and more workplaces rely on
running Oracle by providing them with a SAP R/3
total solution that provides high availability
and scalability. Our team includes highly • More Flexibility
skilled and experienced Oracle DBA’s who are – Dynamical growth by adding servers
trained and certified in SAP basis administra- – Better reuse of equipment
tion.
• Cost Savings
Oracle9i RAC is the highly scalable and – Buy cheaper servers (scale out)
reliable database cluster for mission critical – Reduce maintenance costs
SAP solutions. Our commitment and focus
on this technology enables Delta Consulting Realize Real ROI…the Rewards of Innovation.
to bring the best practices for migrating an Let Delta Consulting provide your organization
existing SAP R/3 4.6D installation from a with the infrastructure services required to
single instance Oracle database to a multiple power your SAP environment.
instance Oracle RAC cluster database confi-
guration.
26

"RMAN has greatly improved reliability of backups and database copies for our customers.
We can now consistently deliver QA and development environments to our customers to
meet their project needs. With automated database duplication, RMAN allows us to perform
trouble free cloning.”
— R I C H B E R N AT,
Sr DBA/SAP Basis Administrator
ChrevronTexaco

CHEVRONTEXACO: ORACLE RECOVERY MANAGER


( R M A N ) D U P L I C AT E – D B A T I M E S AV E R T O T H E
RESCUE

Overview These databases manage critical information


ChevronTexaco relies on Oracle Recovery for all of ChevronTexaco’s SAP modules, in-
Manager (RMAN) for guaranteed backup cluding Financials, HR, Production Revenue
and more importantly, recovery of its Oracle Accounting, BW, and Sales & Distribution.
ChevronTexaco Corp.
databases that underpin a comprehensive suite The databases sustain high transaction rates,
of SAP applications, servicing 25,000+ US- with approximately 3-5% of data changing
Industry:
based employees. In addition, ChevronTexaco on a daily basis, generating 20-50 GB of
• Ranks among world’s
utilizes RMAN to automate the cloning of archived logs daily.
largest global energy
companies
SAP databases for QA and development pur-
poses, and to effectively restore all databases RMAN improves ability to clone database
• Ranks among world’s
largest crude oil and
at a secondary site in the event the production Subsets of the databases are regularly cloned
natural gas producers data center suffers a complete disaster or (or, “refreshed”) for QA and development
• Markets Chevron,Texaco,
outage. environments – in total, 20 TB across 30+
Caltex fuel products databases. Production cloning refreshes occur
worldwide monthly and on-demand. On occasion, project
Introduction
ChevronTexaco Corp. was formed by the database refreshes are requested to different
Annual Revenue:
merger of Chevron Corp. and Texaco Inc. It points-in-time, to accommodate various
$120 billion project phases. In just one day, these refreshes
ranks among the world’s largest and most
competitive global energy companies, with a can incur 6+TB of data movement.
Employees:
125 year legacy tracing back to a seminal oil
50,000 in 180 + countries ChevronTexaco faced many challenges to sup-
discovery in 1879 near Los Angeles, California.
ChevronTexaco is active in more than 180 port ongoing development projects. The SAP
countries, and is engaged in every aspect of DBA group depended on a myriad of SQL and
the oil and gas industry, including exploration shell scripts to perform refreshes. As data vol-
and production; refining, marketing and ume steadily grew, refreshes were not complet-
transportation; chemicals manufacturing and ing within the required refresh window, which
sales. Its fuel products are marketed world- interfered with development schedules. These
wide through the Chevron, Texaco and Caltex challenges, stemming from reliance on OS-
brands. level backup and restore procedures, included:
• Refresh could not be performed to the same
ChevronTexaco’s U.S. SAP DBA group is host, and required a separate database server
responsible for six production databases, • Renaming of Oracle datafiles and directories
encompassing a total of 7.5 TB, accessed by during restore
over 20,000 selfservice users, 3,000 named
• Parsing SQL to rename datafiles and rebuil-
users, and 5,000 financials application users.
ding controlfile during restore
C h e v r o n Te x a c o , R M A N f o r S A P 27

Because of these challenges and the impact on • Development-mandated refresh windows. Backup & Recovery Challenge
internal applications responsible for sensitive For example, a 2 TB refresh to be completed • Creation of QA and develop-
data ranging from employee personal informa- in 9 hours. ment databases not complet-
tion to payroll statements, it became apparent ing within set timeframes,
that the SAP DBA group needed a simpler, Conclusion impacting testing and develop-
more effective backup, recovery, and refresh ment schedules
ChevronTexaco uses Oracle Recovery Manager
method. After investigating several alternative to provide simple and reliable backup, reco-
Backup & Recovery Solution
tools and their associated costs, the IT group very, and database refreshes, while adding no
• Oracle Recovery Manager,
proceeded to standardize on RMAN for all SAP additional license and integration cost to their
Oracle 9.2.0.4
databases. RMAN has met those challenges existing environment. • 6 production SAP databases
head-on. The SAP DBA group reaped considerable cost • 500GB to 2TB per database
savings and productivity benefits: • 250 – 1400 datafiles per
With RMAN, ChevronTexaco consolidated to • Hardware savings through ability to refresh database
a single set of RMAN scripts accessible by all database to the same host, without requiring • 25,000+ named users
databases. By using the RMAN DUPLICATE • Weekly incremental level 0 of
a separate server production databases to tape
to clone a database, ChevronTexaco greatly
simplified their scripts and expanded their level • Tape storage savings using RMAN incre- • Daily incremental level 1
mental backups backups
of automation. • 20 TB across 30+ databases for
RMAN enables ChevronTexaco to: • No additional cost, as RMAN is integrated QA, development environments
with the database • Monthly RMAN DUPLICATE to
• Easily refresh an instance to the same or a
different host, at a current or point-in-time, • Simplified scripts and greater levels of refresh QA instances
automation • Disaster recovery time objec-
and automate renaming of datafiles tive of 72 hrs, for restoring
• Easier delivery of development environment production databases at
• Reduce tape consumption by 80% by
refreshes to customers, meeting their time- remote site
utilizing incremental backups, which only
frames
backup changed blocks
System & Network
Detect physical block-level corruptions • Automated and reliable solution to provide
• Configuration
during backup and restore guaranteed data restoration
• (13) HP N-class & RP servers
• Validate block integrity of weekly full for production, QA, and deve-
RMAN Benefits lopment environments
backups using RESTORE DATABASE
VALIDATE • Guaranteed, accurate backup and recovery • 4-8 CPUs per server

• Ensure that all files needed for restore are • Automatic block corruption detection • 6-20 GB memory per server

present and repair • HP-UX 64-bit

• Configure backup retention policy to easily • Performance-optimized, spacesaving • Oracle9i Database, R 9.2.0.4

obsolete backups that are no longer required backup and restore operations • 1-10 databases/server

• Centralize backup and recovery management • Fine-granular data operations at table- • SAN attached StorageTek

for 40+ databases space, datafile, archive log, controlfile, 9840A and 9980 tape drives,
and block level 1GB fibre
RMAN enables the DBAs at ChevronTexaco • HP OpenView Storage Data
to meet their SLA: • One-step database cloning
Protector 5.1
• Recovery time objective (RTO) for complete • Extensible to third party media managers
restore of the production databases of 72 • No additional license cost – RMAN is an
hours, from declaration of disaster. For integrated feature of the Oracle Database
restoring individual databases, an RTO of 1 Server
hour for every 100 GB of data to be restored.
• Weekly full backup to tape completing in For More Information
5 hrs for a 2 TB production database, with • Oracle9i Recovery Manager User’s Guide
average throughput of 110 MB/sec using 2 • RMAN on OTN
tape devices.
28

“This complex SAP migration to HP Integrity servers ran seamlessly and the result is a trib-
ute to the spirit of the cooperation that existed between Burda, HP, Oracle and Intel. We
gained valuable knowledge, service quality and time.”

— M AT H I A S E N D ,
Manager of System Technology, Burda Digital GmbH

B U R D A D I G I TA L S M O O T H LY M I G R AT E S
SAP ENVIRONMENT TO HP’S POWERFUL
I N T E G R I T Y S E RV E R P L AT F O R M

Burda Digital GmbH End of the road


Hamburg-based Burda Digital GmbH is a “Our SAP R/3 environment had been running
wholly owned subsidiary of the Burda Group. on two AlphaServer GS160 systems on the
Although it has expanded to serve wider mar- Tru64 UNIX operating system and TruClus-
kets, its principal role is still to provide IT ter,” said Mathias End, manager of system
services for all the Burda publishing and print technology at Burda Digital.
businesses. “Unfortunately, the platform had reached the
end of its lifecycle and could not be extended
One of SAP’s first customers, Burda Digital further.”
has decades of experience with SAP business
solutions. Burda Digital is renowned for its Burda Digital decided to seek a new, future-
abilities in this field and is responsible for oriented hardware solution that could improve
providing SAP® R/3® services to 1,300 of existing performance and enable growth.
the 7,400 employees of Hubert Burda Media,
ensuring that the IT environment is capable To find the best solution provider, Burda
of handling up to 800 users simultaneously. Digital assessed contenders on their ability
to deliver technology as well as their ability
To support Hubert Burda Media, Burda to provide an innovative solution. Critical
Digital runs many functions including salary considerations were the planned development
payments, monthly finance calculations, invoice of microprocessor technology, the vendor’s
processing, advertising and sales management experience wimplex SAP R/3 environments
plus a wide variety of other daily business tasks. and the ability to deliver quality service.
The analysis also considered performance and
To handle the various business tasks, fifteen scalability as well as special capabilities such
separate SAP R/3 instances were deployed in a as multi-OS support and partitioning.
three-tier architecture environment, of which On all counts, Burda Digital decided that
six were productive systems running SAP R/3 only one system could be seriously considered:
4.6C. An additional two SAP Business Infor- the HP Integrity server with Intel® Itanium®
mation Warehouse (SAP BW) instances were 2 processors.
also implemented in a two-tier architecture to
handle the data warehouse operations.
B u r d a , O r a c l e t o O r a c l e M i g r a t i o n ( Tr u 6 4 - H P U X ) 29

Proven for SAP Developed jointly by HP and Oracle, the HP


“Our objective was to identify a platform that Smooth Transition Process minimizes transfer
was ideal for all our applications, including times by using procedures already implemen-
data warehousing and CRM, as well as the ted within SAP environments but which,
subscription management system that had been until now, had not been used during standard
developed over many years and on various OS SAP migration procedures.
platforms,” said Thomas Giessler, manager of According to Geissler, it was a “major factor”
system technology applications at Burda Digi- in keeping the project on schedule.
tal. “For this reason, the multi-OS capability
of the HP Integrity platform was closely ex- By choosing the HP Smooth Transition meth-
amined. Another factor in favor of HP Integrity ods rather than more traditional SAP data
servers was the fact that the system is already transfer, planned production downtime was
widely used for SAP and HP could provide kept to a minimum and more SAP systems
many verifiable references.” were migrated within the same timeframe.

HP’s flexibility was another critical factor in The changes for Burda Digital have been
Burda’s decision. According to End, at every profound. The workload previously handled
stage of the project, from initial offer and by eight Alpha processors is now performed by
drafting of the contract, to readiness to support just four Itanium 2-based HP Integrity servers.
ISVs, HP clearly stood out from its competi- The company has substantially reduced its IT
tors. “HP was always willing to deal with our administration overheads and has also made
requirements,” he said. great advances in scalability and flexibility.
Initial tests showed that the new SAP R/3 en-
The solution provided for Burda Digital’s vironment with only four CPUs still delivers
SAP R/3 environment consists of two HP In- an outstanding response of 200 to 400 ms.
tegrity rx8620 servers based on Intel® Itani- Most importantly, Burda Digital has greatly
um® 2 processors, both with 48 GB RAM and increased its customer satisfaction with quick,
running HP-UX 11i. They are connected to an efficient and flexible responses to user requests.
HP StorageWorks EVA 5000 storage system
with a total capacity of 12 TB. This is in turn “This complex SAP migration to HP Integrity
connected to the company’s wider storage servers ran seamlessly and the result is a tribute
network by HP StorageWorks SAN switches. to the spirit of the cooperation that existed
Backup is provided by an HP StorageWorks between Burda, HP, Oracle and Intel,“ added
ESL 9595 tape library, managed by a Veritas End. “We gained valuable knowledge, service
NetBackup Enterprise Server 5.1 running on quality and time.”
an HP Integrity rx2600 server. An additional
HP Integrity rx2600 server is used for test and
quality control in the storage and backup
environment.

Smooth transition
Working to a pre-defined plan, migrating to
the HP Integrity platform was seamless and
efficient and was achieved in less than 36 hours,
due largely to the HP Smooth Transition Process.
This process speeds up and simplifies the trans-
fer of Oracle databases from Tru64 UNIX to
HP-UX 11i.
30

Challenges Solution Results

• It is the responsibility of Burda Digital • Processing power has doubled with just • It was decided to migrate the SAP environ-
GmbH to provide vital SAP business four HP Integrity servers now doing the ment to two powerful Intel® Itanium®
applications to 1,300 employees in virtually work of the previous eight Alpha pro- 2-based HP Integrity rx8620 servers.
every discipline of the Burda Group. cessors and SAP R/3 performance has
• Storage and backup were provided with
accelerated as a result of database reorgani-
• Burda’s SAP environment was running HP StorageWorks EVA 5000, HP Storage-
zation.
on two AlphaServer GS160 systems but Works ESL9595 and two HP Integrity
they had reached the end of their life. • The complex migration ran seamlessly rx2600 servers.
and was achieved in under 36 hours.
• This was compromising not only the • The new HP Smooth Transition Process
current ability to deliver SAP but also • Burda Group now has the business advanta- developed jointly by HP and Oracle was
future growth plans of the group. ges of scalability, flexibility, increased used to speed up the migration of the
customer satisfaction and reduced over- SAPR/3data.
heads.

Technology highlights: HP Services

• HP 24 x 7 support and guaranteed four hour response


Hardware
• HP Proactive Essential Service
• 2 x Itanium 2-based HP Integrity rx8620 servers
• 2 x Itanium 2-based HP Integrity rx2600 servers Why HP?
• On performance and scalability as well as special
• 1 x HP StorageWorks EVA 5000 storage system
capabilities such as multi-OS and partitioning,
with 12 TB capacity
Burda Digital decided that HP Integrity servers were
• HP StorageWorks SAN switches the only systems that could be seriously considered.
• HP StorageWorks ESL 9595 tape library • HP’s Integrity offering is a proven platform for
running the mySAP Business Suite applications and
Software HP has the best knowledge and experience for
implementing SAP.
• SAP R/3 4.6C and SAP Business Information
Warehouse (SAP BW) • HP’s flexibility and willingness to solve problems
made it stand out from the competitors.
• HP-UX 11i v2 rel. 11.23
• Oracle database rel. 9.2.04 version
• Veritas NetBackup Enterprise Server 5.1
COOP Switzerland, Oracle Advanced Customer Services 31

“Databases are important pillars of support when using SAP. If the database systems do
not run optimally or are not available, this will have serious consequences for business
continuity. With Oracle Advanced Customer Service (ACS), we can benefit from an excellent
database support which has paid off for our business in many diverse ways.”

— MANUEL HONEGGER,
Head of Development, Member of the Management.

COOP ENSURES AGREED SERVICE LEVELS


D U R I N G S A P O P E R AT I O N W I T H O R A C L E
ADVANCED CUSTOMER SERVICES (ACS)

The best possible service support for Oracle ACS ensures support expertise and Industry:
Commerce/Retail trade
Oracle SAP customers support quality
Turnover:
Although hundreds and thousands of custo- In addition to this, the Coop has been using Over 14 billion CHF
mers of the retail giant Coop in Switzerland the Oracle Advanced Customer Service for SAP
Employees:
are not aware of the IT system in their daily since the spring of 2003, allowing it to utilize Around 50,000
encounters, its effective functioning ensures a whole series customized support/service
that customers can always obtain the goods Oracle products & services:
options.
• Oracle databases
and products they are looking for at the right Extension of the existing SAP landscape around • Oracle Advanced
time and in the right amount in over 1500 the SAP BW solution was a particularly crucial Customer Service (ACS)
markets. The essential SAP corporate solutions factor behind this decision. Both the require- for SAP
and special commercial applications form the ments for the database systems and the associa- Key Benefits:
basis of a successful business for the second ted support needs increase in conjunction with • Immediate and efficient
largest retail group in Switzerland. this. escalation management
(troubleshooting)
If the SAP Retail and SAP Business Warehouse Although it already has a team of database ad- • Faster upgrades
(BW) systems (all based on Oracle-databases) ministrators, the Coop felt that it was important
are not running, the business continuity, turn- • Reduced downtimes/-
to underpin the above SAP extension with a
over and image of the Coop will be at stake. Maintenance of business
more specialized and intensive database support
continuity
“Our Oracle databases in SAP operations have on a lasting basis. This was primarily intended
always been fundamentally critical to our to ensure that definite service support levels in • Planning for support
success. If they failed or did not run optimally, capacities
respect to system availability, performance and
we would encounter significant problems,” system stability can be adhered to better. • Optimized system
states Manuel Honegger, Head of Development management
at the Coop Switzerland, drawing attention to • Proactive system
As IT Manager Honegger declares, “we require
the high priority of the databases used in the monitoring
support know-how and excellent service quali-
SAP environment. The famous retail group
ty which we ourselves do not have. We there- • Increased productivity in
from Switzerland had “deliberately decidedly
fore decided in favor of the Oracle ACS /SAP the database administra-
for Oracle as part of a strategic evaluation.”
service offer.” The Coop has over 3500 SAP tion
users altogether. In the SAP Retail environ- • Permanent support for
ment, the Oracle database reveals a capability
of more than 4 terabytes, while in the SAP BW
area the 2 terabyte limit has recently been
exceeded.
32

Customized support services are provided Oracle Advanced Customer Support:


As for every Oracle ACS customer, individual Tangible benefits
database services were provided for the Coop Coop IT Manager Honegger provides a
with clear assignment areas. In the case of positive summary for the Oracle Advanced
the Coop, the following Advanced Customer Customer Service implemented: “Oracle ACS
Services modules, amongst others, were pro- for SAP is very clearly worthwhile. We bene-
vided: fit from the services provided as well as from
- Fixed allocation of an Oracle SAP support the expertise and enormous pool of experience
specialist which Oracle offers with ACS Services for
- Guarantee of optimum system performance SAP customers.”
- Support in minimizing the system downtime
- Advice and assistance for system upgrades He continues, “two escalations occurred which
we were successfully able to overcome in a
- Optimum system handling for increasing
very short time with the help of the Oracle
datavolumes
SAP specialist. Without this support, we would
- Know-how and technology transfer in have been confronted with massive problems
Oracle/SAP matters in at least one case and the business continuity
- Implementation and realization of best could have been seriously at stake. Thanks to
practices Oracle ACS, we were able to fulfill the service
“After our positive levels specified.”
- Proactive system management
experience, we no
longer want to do
- Optimization of the total cost of ownership.
The fact that Oracle ACS for SAP enables the
without the Oracle option of a real-time, fully up-to-date exper-
The basis for this is a customized support
Advanced Customer tise/innovation transfer is regarded as a signi-
plan oriented towards the interests of the Coop
Service for SAP.” ficant advantage. Continuous contact with the
with agreed capacity provision. Target/Actual
Oracle SAP specialist allows the Coop to con-
comparisons are regularly performed, as is the
tinuously optimize the Oracle SAP system
exchange of all relevant information over the
operation. “He knows just what is important
remaining time realization of maintenance
– MANUEL HONEGGER to us, and has a profound level of experience
activities, changes to the basic system settings,
Head of Development as well as tremendous knowledge regarding
Member of the Management upgrade changes, Oracle technology renewals
innovations.
Coop AG as well as via the capacities already utilized.
The Coop is able to use a special Oracle Web-
This results in recommendations which allow
site (portal) which serves as a type of infor-
us to maintain our pioneering role and enable
mation and communication platform between
us to continually realize leading-edge solu-
Oracle and Coop.
tions,” emphasizes Honegger, and concludes
by saying “after our positive experience,
we no longer want to do without the Oracle
Advanced Customer Service for SAP.”
34

U S I N G P E R F O R M A N C E A N A LY S I S T O F I N E T U N E
T H E O R A C L E D A T A B A S E S U P P O R T I N G S A P.
R E D U C E D I S C I / O T O A M A X I M U M O F 1 / 1 0 0TH!

Japan Aviation Electronics Industry, Ltd.


Information System Division

Electronic parts manufacturer Japan


Aviation Electronics Industry, Ltd. uses
SAP R/3 to integrate the backend systems
Japan Aviation Electronics
of 18 domestic and international compa-
Industry, Ltd. (JAE)
nies to strengthen their global business.
Headquarters: As usage increased and data expanded,
1-21-2 Dougenzaka, the slowing of response led to a "Perfor-
Shibuya-ku, Tokyo
mance Analysis” in October of 2004 based
Founded:
on the Oracle database solution support,
August 20, 1953
Capital: “Advanced Product Services for SAP”,
10billion 690million yen offered by Oracle Corporation Japan for
Sales: SAP Systems. Not only did this improve
“We named the system ‘J/1’ because it was a
Consolidated 115billion project to make JAE number 1 in all fields.
performance, but it was also successful
800million yen/Independent
96billion 600million (2004) in eliminating error in batch processing. By integrating the backend systems of sales,
Employees: Professional performance tuning services production/supplying, inventory/distribution
Consolidated 4,960/
are excellent supporters in making the
and accounting to a global standard, we have
Independent 1,783 pursued a global operation where associate
(as of March 2005) usage of R/3 more reliable. companies around the world can cooperate
proficiently,” says JAE Information System
Business Summary Strengthening Global Business by
Division’s System Senior Manager Kunio
Focusing on connectors simultaneously introducing SAP R/3
Uehara.
for print foundations and into 18 associated companies
automobiles, JAE is also
expanding to demands in
Japan Aviation Electronics Industry, Ltd. (JAE) Since the company headquarters began full
portable electronics such
started 52 years ago as a company repairing operation in November 1998, group develop-
as laptops and cell phones. aeronautical electronics. Currently, JAE manu- ment has gone smoothly and in August 2003,
Also manufacture system factures four main product groups including the backend systems of the headquarters and
equipment like displays, connectors, system devices such as switches all 18 associated companies were replaced with
Aeronautic/-Astronautic and input devices, aeronautic/astronautic prod- SAP R/3. Unified management of all corporate
products, such as gyros and ucts such as the world-class inertia sensor used activities was made possible by ERP. Oracle
optical devices. Particularly, in the H2A Rocket and optical devices such as was chosen as the central database for this
JAE’s aeronautic/-astronautic optical fiber couplers. Global development has system.
products are adopted in the
been an early focus for JAE, and international “We first chose the Windows technology, and
H2A Rocket and possess
sales accounts for 35.2% of their total sales. then chose Oracle as the most secure and relia-
top level technology
in the world.
Also, as opposed to 7 associated companies ble database in the Windows environment.
domestically, JAE possesses 11 associated The fact that Oracle was the most commonly
companies in Asia, United States, and Great used database for SAP R/3 users in Windows
Britain. environments or in environments that include
JAE’s SAP R/3 based backend system is called programs such as UNIX was also a factor in
“J/1 System.” our choice,” says Mr. Uehara.
J a p a n Av i a t i o n E l e c t r o n i c s ( J A E ) u s i n g O r a c l e C u s t o m e r S e r v i c e s f o r S A P 35

Currently, JAE’s sales are steadily increasing. administrator no longer has to respond to
Total sales in 2004 reached 1.4 times the total problems at night or on their off days. By fol-
sales in 1998. On the other hand, if the inven- lowing the suggestions and revising the logic
tory assets in 1998 were to be 100, they have in the SQL string, it was found that the batch
been condensed to 54 in 2004. processing that took approximately 5 minutes
could be shortened by almost 1 minute. Also,
As usage expands and data increases, there were some SQL strings that reduced the
response is slowed down database’s occupying buffer by up to 1/100th
just by creating a secondary search.
Implementation has achieved great results but
“Overall I/O was reduced and buffers could
it also had problems. Data is growing at a pace
be used more efficiently, making the usage
much faster than anticipated.
of the database easier. Overall performance
“In 2004 when our 18 companies simultane-
should have increased also,” says Mr. Uehara.
ously upgraded to standardize the R/3 version,
“It gave us hints for improvement in the future,
the increase in data accelerated even further.
and taught us a lot. We are beginning to see
A project to archive the data was launched,
the plan to prevent performance decrease,”
but the rate of data increase was greater than “We have more space
adds Mr. Yasuhara.
what could be deleted. Every month data in our database,
Performance Analysis is effective when done
grew by 20GB, and claims from users about and performance has
periodically, 1~2 times a year. If the diagnosis
slow response began to be a problem”, says gone up.”
is performed at times of addition of application
Information System Division’s Kazuo Yasuhara.
modules or storage increases, it can also prevent
It was then that they learned of Oracle — KUNIO UEHARA
problems prematurely.
Corporation Japan’s performance tuning Japan Aviation
“With” Opening Dreams and the Future with
service for SAP users, “Performance Analysis.” Electronics Industry, Ltd
Technology” as their unified concept, Japan
Information System Division System
Aviation Electronics Industry continues to
As a result of Performance Analysis, the strengthen their global operations. In order to
Senior Manager

usage of the database became easier maintain their ERP and Database in optimal
“Performance Analysis” has two phases, a condition, JAE will continue to use “Perfor-
“Performance Diagnosis” and “Tuning” based mance Analysis” in many different ways.
on the result of the diagnosis. JAE performed
a performance diagnosis in October 2004.
A word from the professional
As apparent by the fact that they were success-
ful in the big bang integration of SAP R/3 Issues with database operation such as
without the consultation of an outside party, decrease in performance, increase in data
JAE possessed high enough technology within and needs for upgrades are not foreign to
the company to execute the actual tuning SAP systems. By utilizing our knowledge
operation. The “Performance Diagnosis” did gained through SAP development support, “Oracle gave us hints
not require the stopping of everyday services or for improvement in
we will see that the Oracle used in your
to install a special program, or to even spend
SAP system is operating to fit your needs. the future, and taught
time for preparation. A professional from Oracle
visited the site and analyzed the 12 telebytes If you are wondering how to operate and us a lot.”
of raw data over two days. The report produced maintain your database in SAP, please give
our service a try. — K A Z U O YA S U H A R A
was full of important suggestions.
Japan Aviation
For example, by utilizing Oracle’s new UNDO- — EISUKE SEKIGUCHI Electronics Industry, Ltd
domain, batch processing errors due to ORA- Information System Division
Oracle Corporation Japan
1555 that had been occurring over 50 times a Support Service Headquarter
month was completely eliminated, and the
36

C O N F I G U R I N G A H I G H LY AVA I L A B L E L I N U X C L U S T E R
FOR SAP SERVICES

Clusters of Dell™ PowerEdge™ servers using Oracle9i Real Application Clusters (RAC) can provide SAP ®
software environments with a flexible, scalable, and highly available database platform. The database
will continue to run if one of the Oracle9i RAC database nodes fails; however, vital SAP functionality
such as the message server and the enqueue server can still be single points of failure. To protect
these services from failure and thus unwanted downtime or even data loss, IT administrators can run-
them on a Red Hat ® Enterprise Linux ® OS–based cluster to compliment the Oracle9i RAC database
cluster and maintain service in a highly available manner.

– B Y D A V I D D E T W E I L E R , A C H I M L E R N H A R D , F L O R E N Z K L E Y,
T H O R S T E N S TA E R K , A N D W O L F G A N G T R E N K L E

Setting up a highly available SAP system on Linux Setting up the Red Hat cluster
requires eliminating any possible single point of To set up the Red Hat cluster for the SAP software,
Related Categories: failure for the database as well as for the various SAP administrators should first determine whether the
components of the overall system. While the database Red Hat Package Manager (RPM™) packages for the
Clustering is made highly available by means of Oracle9i Real Red Hat Cluster Suite are installed (see Figure 1).
Database Application Clusters (RAC) technology, SAP applica- Depending on the availability of updates, the version
Dell/EMC storage tions can be made highly available by protecting the numbers the availability of updates, the version num-
High availability (HA) SAP central instance – which includes the message bers may differ. Administrators should install the most
Linux.
server and the enqueue server – from failure. In addi- recent version of these packages. They should then
Oracle
tion, SAP management tools require a common shared prepare the shared storage and the network connec-
Red Hat Enterprise Linux
$ORACLE_HOME directory, which requires the Highly tions. Throughout the example scenario used in this
SAP
Available Network File System (HA NFS) service ex- article, the server names ls3219 and ls3220 are used
Visit porting the Oracle executables, the SAP executables, and for the first and second cluster nodes, respectively.
www.dell.com/powersolu- the SAP shared files such as profiles and the /sapglobal
tions directory.
for the complete category
On Linux, the node membership for Oracle9i RAC [root@ls3220 root]# rpm -q clumanager
index.
database nodes is managed by the Oracle cluster clumanager-1.2.16-1
manager (oracm), which is designed specifically to [root@ls3220 root]# rpm -q redhat-config-
manage RAC nodes. Therefore, administrators should cluster
implement a second, independent cluster to make the redhat-config-cluster-1.0.2-2.0
$ORACLE_HOME directory and the SAP central
instance services highly available. This must be perfor-
med on a second set of hosts, because each node can Figure 1: Checking for Red Hat Cluster Suite RPM packages
be a member of only one cluster. Membership in two
independent clusters with potential conflicts on current
node status would render the cluster nodes unusable for
each of the respective clusters. This second cluster uses
the Red Hat Cluster Suite.
Dell-Linux SAP on RAC - Promotion 37

A Cluster File System allows all nodes in a files on a shared disk, but also store Oracle
cluster to concurrently access a device via the binaries and configuration files (shared Oracle
standard file system interface. This allows for home) making management of RAC even easier.
easy management of applications that need Also any other non-Oracle binaries or non-Oracle
to run across a cluster. OCFS (Release 1) was configuration files (such as shared SAP direc-
released in December 2002 to enable Oracle tories) can now be stored on shared disks.
Real Application Cluster (RAC) users to run the OCFS2 contains the following additional features:
clustered database without having to deal with - Meta data caching
RAW devices. The file system was designed to
- Meta data journaling
store database related files, such as data files,
- Cross node file data consistency
control files, redo logs, archive logs, etc.
- Easy to administer, including operation as
shared root fs
OCFS2 is the next generation of the Oracle
Cluster File System. OCFS2 is available since - Multiple block size support
August 2005 for Red Hat Enterprise Linux 4 - Supports up to 254 cluster nodes
and SUSE Linux Enterprise Server 9 (by default - Context-dependent symbolic links (CDSL)
included in every SLES9 SP2 installation). support for node specific local files
- Async io and direct io support for
OCFS2 has been designed to be a high perfor- database files for improved database
mance general-purpose cluster file system. performance
With it, one can store not only database related - Full integration with Linux kernel from 2.6 on.

Configuring the network [root@ls3219 root]# cat /etc/sysconfig/


Both nodes must have two available Ethernet interfaces. network-scripts/ifcfg-eth0
One interface is used for cluster communication between DEVICE=eth0
the two nodes and should be on a private network. The BOOTPROTO=static
other is the publicly visible network interface. Depending IPADDR=10.17.64.25
on the specific requirements of the environment, admini-
NETMASK=255.255.252.0
strators may want to set up four interfaces—two for each
ONBOOT=yes
node—using the Linux kernel bonding mechanism. This
provides a highly available network connection on each TYPE=Ethernet
channel and secures the cluster against failure of one single GATEWAY=10.17.64.1
com-ponent (network interface card, network cabling, or
switch) on the respective communication channel. Figure 2: Example public interface setup file for node ls3219

Administrators should reserve one public IP address for They should configure the interfaces (or virtual interfaces)
each node. In this example scenario, these addresses are with these addresses, either by using the redhat-config-
10.17.64.25 for node ls3219 and 10.17.64.26 for node network program or editing the respective interface setup
ls3220. Administrators should also reserve one private files in /etc/sysconfig/network-scripts. Figure 2 shows
IP address for each node. For dell3219, the private address what the public interface on ls3219 should look like.
is 172.16.42.34; for dell3220, the private address is Administrators should set up all the interfaces on the nodes
172.16.42.35. according to the host names and IP addresses. This is the
Additionally, administrators should reserve three IP addresses same setup principle that is used in the Oracle9i RAC
for the cluster services to be used as virtual IP addresses. cluster: one public and one private IP address per node.
38

Disk /dev/sdb: 314 MB, 314572800 bytes [root@ls3220 root]# devlabel status
64 heads, 32 sectors/track, 300 cylinders brw-rw---- root disk /dev/raw/raw1
Units = cylinders of 2048 * 512 = 1048576 bytes --[RAW]--> /dev/sdb1
Device Boot Start End Blocks Id System brw-rw---- root disk /dev/raw/raw2
/dev/sdb1 1 300 307184 da Non-FS data --[RAW]--> /dev/sdc1
brw-rw---- root disk /dev/homedir ->
Disk /dev/sdc: 314 MB, 314572800 bytes /dev/sdd1
64 heads, 32 sectors/track, 300 cylinders
Units = cylinders of 2048 * 512 = 1048576 bytes
Figure 5: Checking the status of partitions
Device Boot Start End Blocks Id System
/dev/sdc1 1 300 307184 da Non-FS data

[root@ls3220 root]# raw -qa


Figure 3: Example quorum partitions on host /dev/raw/raw1: bound to major 8, minor 17
/dev/raw/raw2: bound to major 8, minor 33

[root@ls3220 root]# devlabel add -s /dev/raw/raw1 Figure 6: Checking the status of partitions
-d /dev/sdb1
RAW: /dev/raw/raw1 -> /dev/sdb1
Added /dev/raw/raw1 to /etc/sysconfig/devlabel
[root@ls3220 root]# /etc/init.d/clumanager status
[root@ls3220 root]# devlabel add -s /dev/raw/raw2 clumembd is stopped
-d /dev/sdc1 cluquorumd is stopped
RAW: /dev/raw/raw2 -> /dev/sdc1
clulockd is stopped
Added /dev/raw/raw2 to /etc/sysconfig/devlabel
clusvcmgrd is stopped

Figure 4: Binding the partition as a raw device Figure 7: Checking the status of partitions

Configuring the shared storage The next step is to create partitions on the quorum and NFS
storage LUNs. For the quorum LUN, one partition is enough.
After testing the network connections, administrators can
Because administrators will bind these partitions as raw devi-
set up the shared storage. The cluster software needs two
ces, they can set the partition type to “da” (non-file-system
small partitions as quorum devices, which should be confi-
data) with the t option of fdisk. Figure 3 shows what the
gured on separate logical units (LUNs) to maximize inde-
quorum partitions would look like on the host in the example
pendence and minimize possible contention. The partitions
scenario. Administrators should create and format partitions
must have a minimum size of 10 MB each. However, the
on the LUNs for the NFS directories.
usual minimum size for a LUN on a Dell/EMC storage
array is 100 MB. The quorum LUNs will be bound later as Next, administrators should create persistent symbolic names
raw devices. for the partitions with devlabel. This program makes the parti-
Additionally, administrators should create one or more LUNs tion device names resilient against device name reordering
to hold the file systems for the data to be exported via the (for example, when the SCSI scan order is different). In the
HA NFS server. They should follow Oracle recommendations example scenario, the persistent symbolic name /dev/homedir
regarding the size for $ORACLE_HOME and SAP recom- is created for the partition used for the NFS export.
mendations for the executables (approximately 300 to 400 Because the quorum disks are raw devices, they must be
MB depending on the kernel version), and they should take bound so as to be available to the kernel. When the special
into account the data that will be stored in the LUNs as well. symbolic name /dev/raw/rawn is used with devlabel, the link
Once the LUNs have been created on the storage system, is created and the partition is bound as a raw device (see
administrators can make them available to the nodes. Figure 4). Note that the identifier changes to “RAW.”
Dell-Linux SAP on RAC – Promotion 39

Administrators can check the result with the devlabel Administrators can click the OK button and go to File>Save
status command (see Figure 5). to save the changes to the cluster configuration.
Administrators can check whether the raw devices are In the next step, administrators can add a device to mount,
bound. As shown in Figure 6, the raw command displays a service IP address, and NFS clients to the service. In the
the major and minor numbers of the bound devices. example scenario, 10.17.64.27 is used as the IP address for
Administrators can check these numbers against the cur- the HA NFS service. To specify a service IP address, admi-
rently assigned block devices (from devlabel status). nistrators should click the Services tab of the Cluster
Once the devlabel settings have been finalized, administra- Configuration Tool. Then they can select the service and
tors can copy the /etc/sysconfig/dev-label file to the same click the Add Child button. Next, administrators select
directory on the other host. Then, they can log in to that “Add Service IP Address” in the popup window and specify
host and issue the devlabel restart command. Admini- an IP address with netmask and broadcast in the next
strators should not try to add raw devices and symbolic names window. Although the netmask and broadcast addresses are
themselves; they should allow devlabel sort out the unique optional, best practices recommend setting them.
IDs collected on the other node to ensure that the same To add the device to be mounted by the service, administra-
physical device is bound under the same symbolic name. tors can click the Services tab of the Cluster Configuration
Tool, select the service, and click the Add Child button.
Configuring a clustered NFS service They can then select “Add Device” and click the OK button.
Once the network and devices are configured, administra- Then they can specify a device special file (in this example,
tors can activate the Red Hat cluster. Logged in as the root /dev/homedir) and a mount point (/sapmnt/clu_export).
user, administrators can check whether the cluster services Each device must have a unique device special file and a
are running (see Figure 7). If the output does not show that unique mount point within the cluster and across service
the cluster services have stopped, administrators should stop boundaries. Administrators can specify a directory from
them by issuing the stop argument to the init-script. Then, which to mount the device in the Mount Point field. This
they can start the redhat-config-cluster program, preferably directory cannot be listed in /etc/fstab because it is automa-
in a Virtual Network Computing (VNC) session. tically mounted by the Red Hat Cluster Manager when the
service is started. Administrators should choose a file
Administrators should begin by setting up the raw devices system type from the FS Type list (ext3 is used in the exam-
for the cluster quorum. In the Cluster Configuration Tool, ple scenario).
administrators should go to Cluster>Shared State to display
the Shared State dialog box. In this box, administrators Administrators can specify options for the device. If the
should enter the names of the two raw devices: /dev/raw/raw1 Options field is left blank, the default mount options (rw,
and /dev/raw/raw2. suid, dev, exec, auto, nouser, and async) are used.2 Admini-
strators can check “Force Unmount” to force any applica-
They should then add the two nodes as members of the clu- tion that has the specified file system mounted to be shut
ster by clicking the Members tab and going to File>New. down prior to disabling or relocating the service (when the
In the dialog box, administrators should enter the name of application is running on the same cluster member that is
the host (ls3219 in this example scenario). They should then running the disabled or relocated service). When finished,
repeat this step for the other cluster member (ls3220). They administrators can click the OK button and go to File>Save
can leave “Enable SW Watchdog” checked – this enables the to save the changes to the /etc/cluster.xml configuration file,
software watchdog timer, which allows a cluster member to and go to File>Quit to exit the Cluster Configuration Tool.
reboot itself if the system hangs.
Next, administrators can name the NFS service by clicking Testing the cluster
the Services tab and then the New button to display the Once the cluster is configured, administrators can begin the
Service dialog box. They should provide a name without first cluster test. They should test the cluster first on one node,
spaces or special characters (this simplifies querying the sta- and then on the other. First, administrators should restart
tus from the command line). In the example scenario, the the redhat-config-cluster program and go to Cluster>Start
HA NFS service name is RAC_NFS_directories. Neither a local cluster daemons. Once the status display shows that
failover domain nor a user script should be assigned. the host has changed from “Unknown” (red) to “Active”
40

(normal color), administrators can enable the service by • /sapmnt/clu_export/readonly (ro, async)
selecting it in the Services window and clicking “Enable.” • /sapmnt/clu_export/read_write (rw, sync)
The service status should change from “Disabled” (red) to • /sapmnt/clu_export/read_write_root (rw, sync,
“Running” (black). On this node, administrators should no_root_squash)
now see the mounted device under the configured mount Even when it is not immediately necessary to create an export
point. If administrators do not see the device, they should that is root-writeable and preserves the user ID (without
check the system log for cluster service error messages. reassigning nfsnobody to root), best practices recommend
After a successful test on one node, administrators can copy configuring the export with these settings – to enable back-
the /etc/cluster.xml file into the same directory on the other ups and quick file distribution among the hosts.
node. Then, they can start the cluster services there, either
with the init script or with the redhat-config-cluster graphical The allowed client and permissions are attributes of the NFS
user interface (GUI). Administrators also should test swit- Export Client object, which is a child of the NFS Export
ching the service between the two cluster hosts. object. Administrators can add the clients again by selecting
“NFS Export” and clicking the Add Child button. When
finished, the NFS service structure should resemble the
Adding clients to the clustered NFS service
structure of the XML file /etc/cluster. xml (see Figure 8).
After testing that the cluster runs properly, administrators
should extend the configuration of the NFS service to export Note: The NFS export is under the control of the Red Hat
one or more directories to the clients. Administrators should cluster, and the directories exported there must not appear
check that the NFS daemon and the portmapper run on both in the /etc/exports file used by the non-clustered NFS
hosts and are configured to start automatically. They should daemon.
execute the following commands on both hosts:
Client-side mount options
/sbin/chkconfig --level 345 nfs on On the client side, the directories are mounted under
/sbin/chkconfig --level 345 portmap on /sapmnt/homedir/readonly following a schema where
/sapmnt/hostname/directory mounts directories exported by
This enables automatic starting in the runlevels 3, 4, and hostname. All clients mount the exported directories there.
5. Administrators can check the result by entering the The /etc/fstab entries for the example scenario are shown
following command: in Figure 9.
/sbin/chkconfig –list service
Adapting the SAP directory structure
The output should look similar to the following: Locally, symbolic links point to the NFS-mounted directories.
[root@ls3220 root]# /sbin/chkconfig --list nfs For example, the SAP instance DVBGS00 would expect the
nfs 0:off 1:off 2:on 3:on 4:on directory structure shown in Figure 10 on its server. The
5:on 6:off directories are located on NFS and can be found in the same
location (that is, with the identical pathname) on every host.
In addition, the output should look similar for the other
The /usr/sap/RAC/SYS directory links to /sapmnt/RAC (see
service. Administrators should perform these steps on both
Figure 11), and the /sapmnt/ RAC directory links to the NFS
hosts.
directories (see Figure 12).
Next, administrators should return to the cluster configura-
In the example scenario, the NFS directories are organized
tion GUI and click the Services tab. They should select the
by system ID (SID) to support more than one SAP system
NFS service and click the collapse/expand indicator, or twistie,
(see Figure 13). The readonly/ and readwrite/ incarnations
on the left to display the contents. Administrators should
of the RAC_sapsystem directory show that the directories
see the service IP address and the service device. They can
used by an SAP system are divided by these attributes, as
then select the device and click the Add Child button. A
shown in Figure 14.
popup window asks for the export directory name. In this
example scenario, everything below /sapmnt/clu_export is
exported, and the following directories are exported with
different access permissions: Configuring SAP central instance services for the cluster
Dell-Linux SAP on RAC - Promotion 41

can contact the enqueue server via the UNIX® Interprocess


Service RAC_NFS_directories Communication (IPC) mechanism; if they are not part of the
central instance, clients contact the enqueue server via the
service_ipaddress
SAP message server. As opposed to all other components of
ipaddress=”172.16.42.60”
netmask=”255.255.255.0”
the SAP system on the application layer, the enqueue server
broadcast=”172.16.42.255” holds a state – an in-memory table of granted locks – that
cannot be recovered gracefully if the service fails. The mes-
device name=”/dev/homedir” sage server, which consequently plays an important role in
mountpoint=”/sapmnt/clu_export” contacting the enqueue server, holds no state; it receives
fstype=”ext3” only incoming connection requests and transfers them to the
forceunmount=”yes” addressee. The message server can be restarted after failure,
with no impact other than delayed communications. The
nfsexport name=”/sapmnt/clu_export/readonly”
enqueue server is a potential single point of failure in an
client name=”172.16.42.0/24”
options=”ro,async”
SAP system, isolated from the seamless failover provided at
the database layer.
nfsexport name=”/sapmnt/clu_export/readwrite” The SAP solution to the enqueue challenge is the stand-alone
client name=”172.16.42.0/24”
enqueue server and the enqueue replication mechanism. With
options=”rw,sync”
these components, the enqueue server runs as a stand-alone
nfsexport name=”/sapmnt/clu_export/readwrite_root” program and can be contacted directly by enqueue clients.
client name=”172.16.42.0/24” Additionally, a second enqueue server – called the enqueue
options=”rw,sync,no_root_squash” replication server – is started; its only task is to maintain a
second copy of the enqueue state table (lock table). Communi-
cating regularly with the enqueue server, the enqueue repli-
Figure 8: NFS service directory structure
cation server keeps its copy of the enqueue table current. If
Configuring SAP central instance services for the cluster the enqueue server fails, it can be restarted on the host where
the enqueue replication server runs. When the enqueue
Access to data in the underlying database of an SAP system
replication server runs. When the enqueue replication server
is synchronized with a special lock system called the SAP
recognizes that the enqueue server has started up, it will
enqueue mechanism. This mechanism serializes access and
transfer its current lock table before exiting.
prevents access from being changed for more than one
The newly started enqueue server can now continue without
requesting party.
losing valuable enqueue state information. Additionally,
The enqueue server usually runs as a service of the SAP cen- OS-level high-availability software makes the enqueue ser-
tral instance. If clients run in the same SAP instance, they ver available via a virtual, clustered IP address, masking the
restart from the clients so that they always connect to the
ls3220:racadm-DVBGS00 > find /usr/sap/RAC/ same IP address.
DVBGS00/* -prune
/usr/sap/RAC/DVBGS00/data Splitting the central instance
/usr/sap/RAC/DVBGS00/log To secure the SAP system’s services in high-availability clus-
/usr/sap/RAC/DVBGS00/sec
ter, administrators must split the traditional central instance
/usr/sap/RAC/DVBGS00/work
into dedicated instances because a large “service block” can
Figure 10: Directory structure for DVBGS00 SAP instance be difficult to monitor. Furthermore, this large block makes

# HA NFS exports
homedir:/sapmnt/clu_export/readonly /sapmnt/homedir/readonly nfs\ hard,intr,nolock,ro,bg 0 0
homedir:/sapmnt/clu_export/readwrite /sapmnt/homedir/readwrite nfs\ hard,intr,sync,bg 0 0
homedir:/sapmnt/clu_export/readwrite_root /sapmnt/homedir/readwrite_root nfs\ hard,intr,sync,bg 0 0

Figure 9: Example /etc/fstab entries


42

ls3220:racadm-DVBGS00 > ls -l /usr/sap/RAC/SYS/


total 4
drwxr-xr-x 2 racadm sapsys 4096 Nov 11 12:44 exe
lrwxrwxrwx 1 racadm sapsys 18 Jan 24 14:48 global -> /sapmnt/RAC/global
lrwxrwxrwx 1 racadm sapsys 19 Jan 24 14:49 profile -> /sapmnt/RAC/profile
Figure 10: Links to/sapmnt/RAC

ls3220:racadm-DVBGS00 > ls -l /sapmnt/RAC/


total 0
lrwxrwxrwx 1 racadm sapsys 49 Jan 24 14:49 exe -> /sapmnt/homedir/readonly/RAC_sapkernel/exe-640-21
lrwxrwxrwx 1 racadm sapsys 46 Jan 24 14:49 global -> /sapmnt/homedir/readwrite/RAC_sapsystem/global
lrwxrwxrwx 1 racadm sapsys 46 Jan 24 14:49 profile -> /sapmnt/homedir/readonly/RAC_sapsystem/profile

Figure 11: Links tothe NFS directories

restarting services difficult,


[root@ls3220 root]# ls -l /sapmnt/clu_export/readonly/RAC_sapsystem/
because administrators must total 8
also restart parts of the central drwxrwxr-x 3 racadm sapsys 4096 Jan 28 03:16 profile
instance that have not failed.
[root@ls3220 root]# ls -l /sapmnt/clu_export/readwrite/RAC_sapsystem/
To run the enqueue server as total 24
a master/slave service, the two drwxrwxr-x 6 racadm sapsys 4096 Jan 26 18:30 DVBGS00
services should always reside drwxrwxr-x 2 racadm sapsys 4096 Jan 28 02:46 global
on different hosts. The mes- drwxr-xr-x 11 racadm sapsys 4096 Nov 11 11:00 trans
sage server is not bound to a
[root@ls3220 root]# ls -l /sapmnt/clu_export/readwrite/RAC_sapsystem/DVBGS00/
particular host. Because these total 16
are the two services that con- drwxrwxr-x 2 racadm sapsys 4096 Jan 28 06:38 data
stitute a central instance, the drwxrwxr-x 2 racadm sapsys 4096 Jan 27 11:01 log
cluster can run only those drwxrwxr-x 2 racadm sapsys 4096 Jan 26 18:53 sec
drwxrwxr-x 2 racadm sapsys 4096 Jan 28 08:36 work
services, and all application
servers must be outside the Figure 14: Directories used by an SAP system
cluster. However, for systems
management purposes, the message server can run together For each instance, administrators must define an instance
with a dialog service, and an application server can reside in profile and a start profile, according to SAP documentation.
the cluster, or close to it. In the example scenario, the enqueue They can start and stop these instances in the same manner
server; the enqueue replication server; the message server; that they start and stop SAP instances installed by default.
and an application instance with dialog, update, batch, and However, to put these instances under the control of the
spool work processes all run as services in the cluster. high-availability cluster, administrators must provide scripts
for the cluster that conform to the UNIX System V init
In the example scenario, the traditional central instance conventions—that is, Bourne shell (bash) scripts that offer
DVEBMGS is split into multiple instances as follows (the a start, stop, and status function.
two numerals at the end of each instance name represent the
system number): For example instance profiles and start profiles, visit Dell
Power Solutions online at www.dell.com/powersolutions.
• DVBGS00: Dialog, update, batch, gateway, and spool
work processes Switching between different user environments
• DM01: Dialog service (for local administration) and To start the SAP services, the programs use the environment
message server of the SAP administrative user. Because those environment
• E02: Enqueue server parameters differ from instance to instance, a simple way to
• R02: Enqueue replication server switch between different environments for the same OS user
Note that the enqueue server and the enqueue replication is desirable. To achieve this switching, administrators can
server must have the same system number (02) separate from adapt the default environment contained in the sapenv.sh
the rest of the instances; otherwise, the takeover of the and dbenv.sh scripts for each instance, and rename the script
enqueue table will fail. sapenv_INSTANCENAME.sh. Then, they can create a script-
Dell-Linux SAP on RAC - Promotion 43

RAC_app_server DVBGS00 ls3216 10.17.64.22


is used in the example cluster because a service order cannot
RAC_message_server DM01 ls3221 10.17.64.27 be defined, nor can services be set in a relationship. Each
RAC_enqueue_server E02 ls3222 10.17.64.28 service is independently monitored and treated without regard
R02 no address. Bound to the public IP address to the other configured services. Because the enqueue server
of the owning member. and the enqueue replication server are dependent and must be
started and stopped on opposite hosts and in a specific order,
Figure 15: Virtual IP addresses and cluster services for SAP administrators must start and stop the enqueue replication
instances
server from inside the sapenserver-SID script.
RAC_app_server /etc/init.d/sapappserver-RAC Next, administrators can enter the scripts as “user scripts”
RAC_message_server /etc/init.d/sapmsgsrv-RAC in the service definitions and configure a check interval, which
RAC_enqueue_server /etc/init.d/sapenserver-RAC typically varies from 30 to 60 seconds. Before transferring
control of the services to the cluster, administrators should
run the scripts manually to test their functionality.
Figure 16: Corresponding init scripts for cluster services

Building a reliable platform for SAP


let – a reusable script element – containing a source statement
Oracle9i RAC for SAP on Linux can provide a stable, flexible,
such as the following:
and scalable environment, provided administrators follow
#!/bin/sh proper planning and installation procedures. By using the SAP
source sapenv_INSTANCENAME.sh enqueue mechanism with Linux, administrators not only
If administrators source this scriptlet, the user has the can protect the database from unplanned downtime, but they
matching parameters for INSTANCE_NAME in the also can set up the SAP environment to avoid disruptions to
environment. This can also be seen in the start() and stop() end users.
functions of the initscripts package, because the scriptlet is
David Detweiler is the Dell SAP Alliance Manager in Europe, the
sourced before executing the command. Middle East, and Africa (EMEA) and a member of the Dell SAP Compe-
tence Center in Walldorf, Germany. The Dell SAP Competence Centers
Integrating the SAP instances as a cluster service help ensure that current and future Dell technologies work together with
Administrators must create virtual IP addresses for all of SAP solutions and provide customers with the architecture, functiona-
the SAP instances, except for the enqueue replication server. lity, reliability, and support expected of mission-critical applications.
The enqueue replication server always runs on the host not
Achim Lernhard has worked at the Dell SAP Competence Center in
owning the enqueue server and attaches itself to the enqueue
Walldorf, Germany, for three years as part of the SAP LinuxLab. He
server (as opposed to the enqueue server trying to contact
assisted the Oracle9i RAC on Linux pilot customer from installation to
it). It can be bound to the public IP address of the respective
productivity and worked on the hardware certifications.
cluster member, even if this means that it changes IP address
with every service relocation. Florenz Kley is a consultant for SAP Technology Infrastructure. He has
worked for five years at the Dell SAP Competence Center in Walldorf,
The DVBGS, DM, and E instance each require a virtual IP
Germany, as part of the SAP LinuxLab. He conducted performance
address, so that they are always present under the same benchmarks to help prove the scalability and performance of Oracle9i
address from outside the cluster. Because DVBGS and DM RAC for SAP on Linux and helped build the architecture for Dell’s
also appear in the instance list (SM51), administrators should Oracle9i RAC on Linux pilot customer.
adapt the instance name to show the host name belonging
to the service IP address, not the currently active cluster Thorsten Staerk is a consultant at the Dell SAP Competence Center in
member IP address. Administrators can do this by setting Walldorf, Germany, as part of the SAP LinuxLab. He has worked exten-
rdisp/myname to virtualhostname_SID_SAPSYSTEM. In this sively on Oracle9i RAC technologies for SAP, researches new SAP techno-
manner, the instance names remain stable after relocation of logies and functionality, and certifies Dell platforms for SAP on Linux.
the service from one cluster member to another. Wolfgang Trenkle is a senior consultant at the Dell SAP Competence
Center in Walldorf, Germany, and is also a member of the Dell EMEA
Administrators should create cluster services for the SAP
Enterprise Solutions Center team in Limerick, Ireland. In addition to
instances and give each service a virtual IP address as a child.
serving as a consultant and supporting proof of concepts, Wolfgang
Figure 15 shows this configuration for the example scenario,
provides training materials and tools to Dell’s global SAP community.
and Figure 16 shows the corresponding init scripts. As shown
in Figure 15, R02 does not have a service. This configuration “Reprinted from Dell Power Solutions, November 2005. Copyright © 2005 Dell Inc. All rights reserved.”
44

IMPLEMENTING SAP/ORACLE RAC ON


E G E N E R A’ S P R O C E S S I N G A R E A N E T W O R K –
E N H A N C E D H I G H AVA I L A B I L I T Y A N D
PERFORMANCE WHILE LOWERING TCO

Oracle’s Real Application Clusters (RAC) offers a powerful, horizontally scalable environ-
ment for SAP applications. Organizations can gain even more advantages by implement-
ing this clustered application tier on the Egenera® BladeFrame® system with its innovative
Processing Area Network (PAN) architecture.

The BladeFrame allows an organization to The system was designed for mission-critical
implement the entire three-tier computing applications such as databases, application
architecture on a single chassis. This ability, servers and transaction processing making it
combined with the PAN virtualization and ideal for SAP and Oracle implementations.
management capabilities, allows organizations
to achieve the high availability, scalability and Oracle RAC and SAP on the Egenera
performance required for these mission-critical
BladeFrame
applications, while significantly lowering
Egenera GmbH The physical features and built-in virtualization
datacenter complexity and total cost of
ownership (TCO). capabilities of the BladeFrame system provide
a natural and optimized synergy with Oracle
RAC and SAP implementations. The combined
Egenera BladeFrame Overview
solution lowers the TCO for large databases
The Egenera BladeFrame system was designed through:
to better solve the multiple pain points of to-
• High Availability
day’s complex enterprise-class datacenters, which
• Scalability
have evolved in the past few decades based on
• Manageability
a classic server architecture that no longer meets
• Price/Performance
current requirements. By integrating processing,
networking and management functionality
High Availability
previously dispersed across server hardware,
data networks and operating systems, Egenera Egenera blades can be easily configured for the
has created a new architecture, the PAN, which high availability needed for mission-critical
consolidates and simplifies the allocation and applications such as mySAP ERP. A single
management of computing power. spare processing blade can be designated as
the failover blade for multiple RAC nodes,
Egenera’s PAN architecture enables utility replacing the need for hot-standby servers and
computing by providing a pool of stateless clustering software. Unlike legacy x86-based
processing nodes (Egenera Processing Blade™ architectures, the BladeFrame replaces a failed
modules which are in a bladed form factor). node with a backup automatically within
These processing nodes contain no disk drives, minutes. When the “new” Processing Blade
network interface cards or host bus adapters: is started, it has exactly the same state – the
nothing that would give the server a specific application with the exact network and stor-
identity or state. age configurations – of the failed blade. The
BladeFrame’s virtualization capabilities trans-
Instead, all points of state and identity have fer all of the server’s characteristics. Thus,
been moved to the fabric switch, enabling the time that system capacity is compromised
virtualization of both servers and connecting due to a failed node is reduced from days or
datacenter infrastructure. hours to minutes.
Oracle RAC and Egenera 45

All of the switching internal to the BladeFrame The BladeFrame also excels at ongoing system
is virtualized, automatically creating high management. Its virtualization technology
availability for network, storage and console enables a customer to move an Oracle RAC
connectivity. database from one Processing Blade configura-
tion to another, or add a node to a cluster, on
Scalability the fly. In addition, Egenera PAN Manager™
Unlike high-end SMP machines, which have software dynamically allocates SAN disks to
finite limits in their ability to add processing servers, optimizing the functions provided by
capability, the BladeFrame can readily deploy Oracle Automated Storage Management,
new processing ability to meet dynamic user which spreads database blocks across multiple
demands. The BladeFrame creates a pool of SAN disks/LUNs. The combination of Oracle
processing blades that are deployed entirely ASM and Egenera PAN Manager software
through management software and without allows users to migrate a live Oracle database
physical intervention. Hence, a single Blade- to a different set of disks/LUNs without any
Frame could quickly scale to 24 Oracle RAC interruption of service or downtime.
nodes, bringing as many as 192 processors
and 768 GB of memory to work against the Price/Performance
database. The BladeFrame’s combination of high-per-
formance x86 CPUs and a dedicated, high-
With the long delays required to bring legacy bandwidth, low-latency backplane provides
servers online, enterprises have had to size their unparalleled price/performance for three-tier
database servers to meet peak demand and high SAP applications with Oracle RAC. The
availability requirements, resulting in utiliza- Egenera 2.5 Gbps high-speed low-latency
tion rates around 20 percent. The BladeFrame backplane allows for faster communications
increases server utilization by correctly scaling among database, application and Web appli-
both the type and number of Processing Blades cations running on the Egenera BladeFrame.
allocated to an SAP application. For example, Oracle databases communicating with other
BladeFrame Processing Blades form a pool of application tiers will experience a higher level
processing capability that can be brought into of performance than if that communication
service when needed and repurposed when no was via a standard Ethernet network. The
longer required. In addition, applications can backplane provides for improved performance
be easily moved from one size of blade to of Oracle databases, by providing a high-speed
another to match the processing requirements. network for inter-node message passing co-
ordinated using Oracle’s Cache Fusion.
Manageability
Deploying a cluster on the Egenera BladeFrame Lower TCO
is very easy. Egenera’s virtualization capabilities Compared to traditionally architected servers,
reduce the number of hardware components the Egenera BladeFrame and its PAN archi-
required for standard server configurations by tecture would save approximately 50 percent
80 percent. There are no NICs, HBAs or SAN in total costs over a three-year period, including
switches to configure and cable. All configu- substantial top-line benefits for Oracle/SAP
ration is done through software. By transfor- installations:
ming manual, physical tasks into automated,
software-based tasks, the BladeFrame allows • Up to 70 percent savings on capital and
system administrators to rapidly establish an maintenance costs
Oracle RAC installation for SAP. • Reduces labor costs for configuration and
ongoing support by up to 80 percent
46

Egenera/SAP/Fujitsu Siemens Computers under the brand PRIMERGY BladeFrame as


Alliance part of their Dynamic Data Centre solutions.
Egenera is a new SAP Global Technology
Partner certified for Microsoft Windows with More information can be found at:
a Competency Center established in Walldorf. www.egenera.com
The innovative Egenera BladeFrame with its www.fujitsu-siemens.com/le/products/stan-
PAN architecture enables a dynamic IT envi- dard_servers/bframe/index.html
ronment that supports SAP NetWeaver and
mySAP mission-critical requirements while Contact:
lowering datacenter TCO and complexity – Egenera GmbH
enabling superior performance. SAP Competence Center
Altrottstr. 31, SAP PartnerPort
Fujitsu Siemens Computers, in a strategic D-69190 Walldorf, Germany
OEM alliance with Egenera, is selling and Phone +49 6227 73-3480
supporting the Egenera BladeFrame in EMEA sapcc@egenera.com

SAP INFO
SOLUTIONS: Active Global Support ISSUE 7 JANUARY 2006

As was the case with Oracle Version 9i, high-speed memory interconnects. As a
W H AT ’ S N E W ? SAP will only certify the terminal release result, Oracle RAC now offers effective
of Oracle 10g (that is, Version 10.2). This new solutions at database level for ensu-
strategy has proved a great success in terms ring 24/7 system availability or adapting
Oracle Version 10g available of quality in the past. For SAP customers, dynamically to growing or uneven system
an added benefit of starting at the terminal loads.
Now that Oracle Version 9i has been on
release is the reduced amount of effort
the market for a few years, the time has SAP has been working on the Oracle RAC
involved in upgrading from other release
come for its successor, Oracle Version 10g project since 2002 and, in collaboration
tracks of 10g. Of course, Version 10.2 also
to be made available to SAP customers. with Oracle, set up a validation process to
boasts a raft of new features, with special
The first step towards that goal was taken certify the operating-system specific use
focus on manageability and database ad-
in the last quarter of 2005, when mySAP of Oracle RAC. The first customer went
ministration.
Business Suite 2005 and its platform live before the end of 2002 with an RAC-
SAP NetWeaver 2004 were launched; Oracle 10g aside, one of Oracle’s hottest based SAP R/3 system running on the
the latter only supports Oracle 10g. The properties for SAP customers at the mo- operating system Tru64. Recently the cus-
validation process for certifying Oracle ment is the parallel database variant RAC tomer pilot projects for the final remaining
10g for use with SAP products based on (Real Application Clusters), which was operating systems were als successfully
the 4.6D_EXT and 6.40 kernels (for already available as an optional extra in completed. Oracle RAC is now officially
example, SAP R/3 4.6C SR1, SAP R/3 Oracle9i. Although its predecessor OPS available to all SAP customers. Around
Enterprise 4.7, SAP NW 2004, and (Oracle Parallel Server) did not prove 10 SAP customers are running a live
mySAP ERP 2004) began immediately adequately scalable in SAP environments, RAC-based system on a variety of opera-
thereafter and will conclude in the first those problems have now been overcome, ting systems.
quarter of 2006. thanks in part to innovations, such as Dr. Christian Graf, SAP AG
Oracle Partner for SAP 47

“The expertise and depth of TUSC allows us to focus on strategic projects…”

– M A R S H A D U V D E VA N I
Manager
Database Administration IFF

INDUSTRY LEADER CHOOSES TUSC


REMOTE DBA

Challenge Solution Results


An acquisition and IFF chose TUSC Remote IFF maintains over
turnover caused a DBA to monitor its 40 databases in
dramatic increase in data- Oracle environments and 7 countries with one
base administration mentor its Oracle DBA. full-time DBA.
and monitoring needs.

About TUSC

IFF is a leading creator and manufacturer of The four-year partnership with TUSC began TUSC is an Oracle Advantage
flavors and fragrances used in a wide variety of when IFF acquired Bush Boake Allen, a for- Partner providing expert-level
consumer products – from fine fragrances and mer competitor and TUSC client. Duvdevani integrated solutions to
toiletries, to soaps, detergents and other house- explained, “One of our DBAs knew of TUSC Global 2000 customers since
1988. TUSC supports SAP
hold products, to beverages and food products. and we needed additional Oracle expertise,
clients running on the Oracle
IFF is dedicated to The Pursuit of Excellence in so it made sense to add our databases to the
database through our Remote
every area of its business, using knowledge, crea- TUSC contract. TUSC enabled a smooth
DBA organization. We have
tivity, innovation and technology to continually transition of incorporating support of the newly supported SAP running on
provide customers with the highest quality acquired system and by mentoring our DBA.” Oracle for many years with
products and service and superior consumer our Remote DBA service while
understanding. IFF has come to depend on Remote DBA and also helping SAP clients
Managed Services, and benefits from TUSC’s in performance tuning and
IFF has sales, manufacturing and creative facili- technical knowledge and expertise. “We benefit other DBA tasks through our
ties in 34 countries worldwide and annual sales from our TUSC DBA’s experience with count- Database Administration
exceeding $1.9 billion. In 2000, IFF acquired less other systems,” Duvdevani said. Through group. TUSC provides a
comprehensive set of con-
Bush Boake Allen, formerly the industry’s 7th a web portal, which includes system reports
sulting services in the areas
biggest player. To manage its Oracle systems and documentation on all work performed, “We
of strategy, architecture,
environment, IFF relies on TUSC Remote DBA. know that our systems are being maintained,”
implementation, support
Duvdevani continued, “The expertise and depth and training.
“IFF has over 40 Oracle databases, including a of TUSC allows us to focus on strategic pro-
large ERP implementation,” stated Marsha jects, not system maintenance, knowledge gaps Get information about TUSC
Duvdevani, Manager, Database Administration. and turnover.” at www.tusc.com
With systems in the United States, Australia, or send an email to:
South Africa, India, Sweden, Turkey and Singa- NIEMIECR@Tusc.com
pore, 24/7 monitoring, performance tuning and Phone +1 630-960-2909
emergency support is essential.
48

ORACLE PREMIUM SERVICES FOR SAP R/3


CUSTOMERS

Oracle is the most widely used platform in SAP R/3 environments. Oracle Services offers
comprehensive services to all SAP customers: Premium core technology services and
workshops, as well as migration and integration services.

Oracle and SAP Core Technology Services for SAP R/3


More than three quarters of all SAP customers Customers
entrust their data to a core Oracle database. Performance analysis and recommendations
Over a million registered users at about ten
thousand locations have come to expect the best Getting optimized performance from the SAP
technology and performance that the Oracle R/3 system is one of the primary objectives of
database brings to the SAP R/3 environment. R/3 customers. A parameter setup for the
Its outstanding attributes are a direct result of customer’s environment cannot be optimized
over ten years of intensive partnership between without specific knowledge of R/3, since the
SAP and Oracle. patterns of R/3 database access are rather dif-
ferent than those of other Oracle implemen-
Naturally, this SAP and Oracle cooperation tations. Therefore the Solution Center’s SAP
does not end with technology exchange. Custo- Support and Services group offers a service in
mer support and satisfaction are the tangible which it monitors the customer’s system over
day to day benefits of this partnership. A key a defined period of time and then recommends
enabler of this customer support is SAP’s parameter settings that would be optimal for
requirement for permanent on-site consulting the customer.
staff to provide immediate support ranging from
application development to critical problem The development of performance-based
response. The SAP Support and Services group ABAP/4® statements is also critical to the
at Oracle’s Solution Center was created to co- database’s performance. Taking advantage of
ordinate this focused customer support. We the performance measurement tools provided
are a worldwide services group providing by the R/3 System can make it very easy to
assistance in all aspects of SAP R/3 operation. analyze even the most expansive statements.
Specifically our mandate covers tasks ranging The Solution Center’s SAP Support and Services
from management of increasing data volumes, group also offers a detailed analysis and fine
performance optimization and setup of high tuning service for the customer’s system
availability solutions, to data exchange bet- specifically from the perspective of the Oracle
ween the SAP R/3 system and other Oracle database.
databases over a secure interface.
Hot standby databases
Our offering of SAP-specific services covers SAP R/3 system downtime often disrupts pro-
the design, operation and management of your ductive activity. In the case of a system crash,
SAP R/3 environment. Our services include core recovery might take several hours during which
technology services, migration and integration time it is impossible to perform any effective
services, as well as a workshop program. work on the system. The duration of system
downtime can be minimized by implementing
We make sure that you stay on the leading a hot standby database. This means that the
edge: Oracle Services in concert with SAP is customer sets up two similar systems. The
leveraging over ten years of experience in first system operates as a normal database server
delivering R/3 solutions founded on the Oracle with all users connected. The second system
database. operates as a backup for the first database.
Oracle PREMIUM Services for SAP R/3 Customers 49

Backup operations continually recover all Migration Services for SAP R/3
incoming archive files from the first database
server. In the event of a crash of the database A great majority of all SAP R/3 installations
server, the ‘hot standby’ server can be opened run on an Oracle database. The Solution
very quickly, and all application servers can Center’s SAP Support and Services group is
connect to that system. Then the administrator certified by SAP for R/3 database migrations
has enough time to analyze the crash while all and can perform migrations for companies who
of the users continue to work. This technique wish to migrate their SAP R/3 installation
is fully supported by SAP for R/3 Release 3.0F from a non-Oracle database to Oracle. Profes-
and Oracle7 Release 7.3.3 or higher, and it has sional service can be provided for migrations
been implemented at a number of customer of any SAP database. Several customer systems
sites. The Solution Center’s SAP Support and have already been successfully migrated by the
Services group offers assistance in implementing Solution Center’s SAP Support and Services
and testing hot standby solutions by using group. We offer you the best migration service
either the SAP-provided tool implementation you can get, because we have the experience
or Oracle tools for Oracle releases prior to 7.3. needed to handle any problems that may arise
from the database administrator’s perspective.
Setup and reorganization
The Solution Center’s SAP Support and
Services group can also offer customers intro-
The constantly increasing data volume of a ductory training in using the new database.
SAP R/3 installation can often no longer be
managed by a standard setup. Parameter
Migration steps
changes need to be specifically targeted to the
Oracle database underlying SAP R/3. Our team • Analysis of current system
designs and implements Very-Large-Database - Determine storage capacity of the
concepts brought in line with the R/3 require- destination system
ments. Where desired, the Solution Center - Examine database-specific configura-
SAP Support and Services can provide help tions and customer-specific R/3
in writing reorganization scripts, so data implementations
restructure can be executed in an acceptable
• Migration to a test system
timeframe for any table structure.
- Write scripts for the necessary
Backup/restore/recovery concepts for mission- unloads
critical systems - Start the database unload
The R/3 System often represents the most - Create space calculation for your
important processes of a company’s business. target database within the database
Downtime therefore quickly leads to disruption unload
of productive activities. Companies need a - Load the unload files to a test system
detailed recovery methodology. The Solution
- Conduct intensive testing of the
Center’s SAP Support and Services group pro-
migrated system (performed by the
vides special backup/restore/recovery works-
customer) over a time period of 2-4
hops in which different backup procedures and
weeks
rules are discussed. They also offer tips and hints
in using the tools provided with your SAP • Migration to the productive system
software. Oracle services can also be requested - Modify the import scripts based on know-
to assist in setting up, implementing and ledge gained from import to the test
testing the backup strategy. system
- Perform the load on the production system
50

• Examination of migration results The workshop is usually held at the customer’s


- A special team within SAP is available to site so that examples can be discussed and
examine the customer system via remote described within the customer’s own environ-
connections ment. The workshop includes discussion of
• Working instructions error recognition and troubleshooting for
business-critical systems.
- Short working instructions are given to Participants receive an extensive documentation
the DBA packet in English on the topics covered. It can
- Explain major administration steps be used as a reference in daily work or as a
- If requested, migration can be followed troubleshooting guide for any problems that
up by an administration workshop. might arise.
The fundamentals of the Oracle database and
SAP R/3 system are covered, with particular
Workshop: Oracle for SAP R/3 emphasis on interaction of the two systems, and
The Workshop: Oracle for SAP R/3 is designed it includes backup/restore/recovery concepts.
for database administrators of an R/3 system
to get, in a compact and practical form, all of Sample agenda for a five-day workshop
the information they need to run an R/3
Day 1
system optimally.
Oracle startup

The workshop is run by professional Oracle Background: Process architecture


consultants whose experience is based on daily Physical Structure of the DB
support of R/3 customers. They are technical Logical structure of the DB
consultants rather than teachers, which means
they have experience in handling the types of Day 2
problems that customers encounter. Customers Review of topics covered
can profit from their experience gained at
Basics of the DB and SAP R/3
other customer sites.
How does Oracle process a transaction?

The workshops have been well-received by Special SAP areas


many customers worldwide. References in Day 3
Europe and US are available upon request.
The workshop contents can be tailored to suit Review of discussed topics

customer needs. Rollback segments


Archiving online redo log files
The workshop is designed to be 5 days in Security of a SAP R/3 System
length. The following agenda is an example that
TSP layout for Oracle/SAP databases
can be modified for each individual customer.
Reorganizations

A typical group of workshop participants con- Day 4


sists of 2-6 administrators who have already
gained some experience on their own systems. Review of topics covered
The course is structured to reinforce lessons Tuning of a SAP R/3 system
already learned. The workshop can be adapted Day 5
to specific customer requirements; the topics to
be covered are discussed with the customer in Review of topics covered
advance and an agenda is created. Backup/Restore/Recovery CONCEPTS
Error handling
Oracle Integration 51

CONNECTING SAP TO YOUR ENTERPRISE


A N D B E Y O N D W I T H O R A C L E I N T E G R AT I O N

Oracle Integration is leading the industry Oracle Integration


with native support for standards. Discover The products that make up the Oracle inte-
how SAP customers can integrate to other gration:
packaged applications, legacy systems and
trading partners easily and effectively.

Staying competitive in today’s ever-changing


environment is key to the success of any busin-
ess. Maintaining a competitive advantage
requires constant innovation. Traditionally,
organizations focused their innovation on new
products or service offerings. More recently,
they have begun to innovate the way existing
products or services, are done by developing
unique business processes. As companies turn
to technology to manage business processes,
and as they integrate disparate systems with
these processes, the ability of their IT infra-
structure to quickly adapt becomes critical. Oracle BPEL Process Manager makes 100%
Organizations need an IT infrastructure that business process portability a reality
works seamlessly to deliver a nimble, efficient,
and highly adaptive set of services, so that Oracle’s latest product offering, the Oracle
business processes can be created, deployed, BPEL Process Manager, lets companies orche-
and optimized with little investment and strate and execute Web services and business
minimal time to value. processes. Oracle BPEL Process Manager is
the first product to deliver native support for
The emergence of service-oriented architectu- the critical Service-Oriented Architecture
res promises to hasten the time required to process standard, which reduces the cost and
integrate disparate systems and to quickly complexity of integration projects while
deploy services through rapid development and increasing their strategic value. Oracle BPEL
component reuse. The ability to capture, corre- Process Manager, based on the Business
late, and aggregate events from many event Process Execution Language for Web Services
sources in a highly integrated environment standard, is a result of Oracle's acquisition of
ensures that your organization will be able to Collaxa Inc.
predict, and respond quickly to, changing Ensuring 100 percent portability and reuse
business conditions. Business process optimiza- of processes, Oracle BPEL Process Manager
tion closes the loop on developing, deploying, delivers a production-level, deployment proven
and improving business processes, enabling solution for implementing the BPEL standard.
your enterprise to improve its business with This proven process standard not only speeds
minimal overhead. the deployment of integration projects while
reducing costs, but enables rapid redeployment
of best practices. Without a standard process
technology, merging or redeploying processes
52

Moving to a Service-Oriented Architecture and customer satisfaction. To improve on-


The BPEL standard, managed by OASIS, is time delivery and lower inventory costs,
the cornerstone of process orchestration and Draeger launched an ambitious logistics
execution within a Service-Oriented Archi- initiative.
tecture (SOA). As many leading organizations Draeger chose option b, which it implemen-
adopt SOAs to reduce development time and ted using Oracle AS Integration and Oracle
costs, BPEL will provide core process orches- RAC, a technology combination that gua-
tration capabilities. Together, SOA and BPEL rantees a uniquely high, availability solution.
deliver a modular, efficient, and effective way According to an IDC Return on Investment
for IT departments to save time and reduce study, this approach saves Draeger about
costs in the deployment of new solutions, $6M on IT-related and implementation costs.
decrease maintenance costs, and quickly adapt In addition, the time to integrate newly
to changing business requirements. acquired companies into the supply chain is
greatly reduced. Altogether, Draeger accom-
Draeger sees massive ROI and lightning plished a 193% ROI through this supply
speed implementation of their global
chain integration project. IDC analyzed that
the payback period for Draeger’s investment
Integration initiative with Oracle AS
in Oracle AS Integration, RAC and PointOut
Integration technology was barely half a year.
Draeger, a $1.3B company with 9,000 em-
ployees, is the world’s top supplier of Safety
and Medical gear represented in over 30
For more Information
countries. By the mid 1990s, acquisitions
created a tremendous challenge for Draeger’s http://otn.oracle.com/bpel
global supply chain. Disparate data warehouses http://www.oracle.com/solutions/integration/
and ERP systems were affecting profit margin index.html
54

USEFUL LINKS ORACLE FOR SAP

www.oracle.com/newsletters/sap SAP OSS Note 23875:


Supported database and operating system
www.sap.com/partners/directories/technology/epx
releases for SAP Kernel version 3.1I^
http://service.sap.com/dbaora-BC-DB-ORA-RAC
SAP OSS Note 720886:
http://service/sap/com-PAM Oracle 10g – Integration in the SAP Environment
SAP OSS Note 581312:
SAP on RAC in general, please see
Licensing
http://service.sap.com/dbaora -> Media Library -> Oracle RAC
SAP OSS Note 740897:
SAP OSS Note 828268: Licensing
Oracle 10g features SAP OSS Note 5740897:
SAP OSS Note 527843: Licensing
10g and RAC for more details SAP OSS Note 581320:
SAP OSS Note 156548: FAQ Oracle RAC for SAP
Supported database and operating system SAP OSS Note 524816:
releases for SAP Kernel version 4.6x Enqueue Service for RAC SAP-Enqueue-Service
SAP OSS Note 85838: SAP OSS Note 705608:
Supported database and operating system SAP RAC specific monitors
releases for SAP Kernel versions 4.5x and 4.0x
SAP OSS note 830982:
SAP OSS Note 407314: HW fine tuning for SAP on RAC environment
Supported database and operating system
releases for SAP Web Application Server 6.x SAP OSS Note 105047:
Partitioning and Oracle Functionality
SAP OSS Note 23875: supported by SAP
Supported database and operating system
releases for SAP Kernel version 3.1I^ SAP OSS Note 701235:
Compressed Data Segments

Oracle Corporation

World Headquarters
500 Oracle Parkway
Redwood Shores,
CA 94065
U.S.A.

Published by
Oracle Corporation,
Oracle for SAP Global
Technology Center
Altrottstr. 31
69190 Walldorf, Germany
Tel. ++49 (0) 6227-8398 - 0
Fax ++49 (0) 6227-8398 - 199 Copyright © 2006 Oracle. All Rights Reserved. Published in the U.S.A. This document is provided for information purposes only, and the con-
tents here of are subject to change without notice. This document is not warranted to be error-free, nor is it subject to any other warranties or
E-Mail: conditions, whether expressed orally or implied in law, including implied warranties and conditions of merchantability or fitness for a particular
saponoracle_de@oracle.com purpose. We specifically disclaim any liability with respect to this document, and no contractual obligations are formed either directly or indi-
Albrecht Haug rectly by this document. This document may not be reproduced or transmitted in any form or by any means, electronic or mechanical, for any
albrecht.haug@oracle.com purpose, without our prior written permission. Oracle is a registered trademark of Oracle Corporation and/or its affiliates. Other namesmay be
Internet:
trademarks of their respective owners. SAP, R/3, mySAP, SAP Business Warehouse and all SAP product names mentioned are used for identi-
http://www.oracle.com/
fication purposes only, and may Copyright©. This document is provided for informational purpose only and the information herein is subject
newsletters/sap
http://www.sap.com/ to change without notice. Please report any errors herein to Oracle (saponoracle_de@oracle.com). Oracle Corporation does not provide any
partner/index.htm warranties covering and specifically disclaims any liability in connection with this document.