Professional Documents
Culture Documents
S A N Qu i c k R e fe re n ce High Availability
SANs offer many options that can provide uninterrupted
SANs are built up from many types of
hardware components. Some of these
This can lead to difficulties when matching
up devices from different vendors and SAN Benefits access to data and applications, even if equipment fails or a
disaster strikes. By incorporating Server Clusters, RAID
devices, such as Fibre Channel switches, suppliers. Much of an individual SAN's
Administrating all of the storage resources in large, Server-free backup, are made possible by a SAN's Storage Systems, Remote Data Replication, and robust
are often unique to SANs. The physical design is concerned with the interoperability
high-growth, mission-critical data centers can be quite flexible architecture and can improve overall Management Software into a SAN, continuous access to
SAN itself frequently includes RAID of its member devices. How these elements
daunting and expensive. SANs represent a way to performance significantly. Even storage reliability can data assets can be achieved. Note that data is only part of
storage systems, hubs, switches, bridges, interconnect, particularly in High Availability
dramatically reduce these costs and complexities be greatly enhanced by special features made the overall availability requirement since applications must
servers, backup devices, interface cards, configurations, may not always be intuitive-
while providing significant technical advantages. They possible within a SAN. Options like redundant I/O also be protected. Therefore, a well implemented SAN
and specialized cabling. These can come even for storage veterans. And, because
can provide large increases in storage performance, paths, server clustering, and run-time data replication should include provisions for application specific
together to form a storage "system," or SANs are typically just as dependent upon
facilitate shared access to equipment and data, (local and/or remote) can ensure data and application monitoring and application fail-over. Additionally, a SAN
network, that provides the storage software for their operation, it can be vital to
greatly simplify scalability, and offer state-of-the-art availability. Adding storage capacity and other should keep data protected from non-hardware errors,
resources and facilitates the policies of an secure the latest software (and firmware) in
reliability. In fact, the I/O performance of a SAN can storage resources can be accomplished easily within such as operating system or database corruption.
IT organization. It is very important to order to avoid potential problems. For these
select and configure these hardware reasons, companies that specialize in be much higher than that of traditional direct-attached a SAN, often without the need to shut down or even
devices with care. A successful SAN supplying complete SAN solutions can be storage, largely because of the very high data transfer quiesce the server(s) or their client networks. These,
implementation needs to be well planned worth seeking out. Their 'real world' rates of the interfaces used to connect devices in a and other, features can quickly add up to big cost
and requires a good understanding of the experience in designing and deploying SANs contemporary SAN (such as Fibre Channel). savings, painless expansion, reduced network
operating principles, and "standards," that can save great amounts of time and money. Additional techniques, such as load balancing and loading, and fewer network outages.
are involved. Some of these standards In many cases, SAN specialists represent
are still evolving and haven't been equally the quickest and most efficient way to
adopted by all equipment manufacturers achieve the many benefits available from
and software developers. Storage Area Networks (SANs).
Remote Mirroring
Server
RAID B'
Remote Replication
Data Replication provides many benefits in today's IT
environments. For example, it can allow system
administrators to create and manage multiple copies of vital
information across a global enterprise. This enables disaster
SAN Admin Console
(Remote)
recovery solutions, maximizes business continuity, and
permits file server content to be distributed over the Internet.
Replication options can even improve host processing
efficiency by moving data sets onto secondary (often remote)
Ser ver Clustering servers for backup operations. In some cases, these data
replication capabilities are required by the "high availability"
In a SAN context, server clustering generally refers to the grouping together of servers for the and "server clustering" features provided by many of today's
purposes of sharing storage resources and data, enhancing their performance (i.e. Load Backup SAN architectures. Remote data replication is typically
balancing), and/or providing fail-over protection in the event that a member server achieved with one of two basic strategies: Storage
misbehaves or malfunctions. Though servers can be clustered together outside of a SAN Solutions replication is focused on the bulk transfer of files, or block
environment, there are several benefits associated with clustering them together as part of a data, from one server to "one or more" other servers. This
common SAN. One of these is the capability to provide uninterrupted access to data, and One of the most valuable time and cost- burden that is usually placed upon a
type of replication generally allows application(s) to be
applications, during and after a server failure. Other benefits often include shared access to saving features of a SAN architecture is dedicated backup server as backups are
running on a server while they, and/or their data, are
disk and tape backup equipment, simplified scalability, data replication options, and increased its ability to offload backup operations performed. This is typically achieved by
being replicated to another off-site server.
I/O performance. In many cases, the specialized software involved in from LANs and servers. This capability embedding some of the backup WAN Application level replication is specific to a
server clustering can even fail-back a server in the event it is repaired can significantly increase the available intelligence into the data storage devices
particular application, such as a database or web
or begins working properly again (often automatically). Some bandwidth on a LAN to network themselves (RAID systems and tape
server, and is typically performed at the transaction
clustering software can even divide application tasks among LAN clients and end users during backup drives) or SAN connectivity peripherals
level (field, row, table, etc.) by the application itself.
clustered servers in order to dramatically improve their response operations. When traditional (switches, hubs or bridges). This can
Many replication products include the ability to
backup servers are relieved from free up traditional backup servers
time and overall performance. Server clusters can also be very transfer data Synchronously or
valuable when considering disaster recovery options. In these "handling" backup data, they can significantly by releasing them from
Asynchronously. With synchronous transfers,
cases, the clustered servers can be joined together with WAN be repurposed and made available data moving duties and large
each packet of transmitted data is
links instead of local SAN connections in order to provide for other tasks. LAN-free backup portions of a backup operation's
acknowledged by the receiving server
enough geographic distance for disaster protection. Locally Clustered solutions can optimize backup administration. When
before more data is sent to it. This can be
Servers operations by offloading backup implemented properly, these
FC Switch FC Switch a slower form of replication, but is very
traffic from a LAN to a SAN, SAN based backup
reliable. Asynchronous data transfers
thereby increasing the amount of solutions let administrators
Remote Cluster Member
FC Switch SAN A SAN A' allow data packets to be sent ahead of
LAN bandwidth. Serverless optimize network and
SAN acknowledgements from the receiving
backup, on the other hand, server utilization,
WAN server for previously sent packets. This
Shared Storage extends these performance gains dramatically shorten backup RAID A RAID A'
method is usually faster but allows more
even further by offloading more than times, and regain processor
RAID C data to be lost if links fail. Eventually, in
RAID C' 90 percent of the administrative and network resources.
Backup Server either case, all transmitted packets must be
Disaster Recovery acknowledged by the receiving system.
For more information call 800-872-2783 more
or visit www.dothill.com websales@dothill.com Copyright 2001 Dot Hill Systems Corp.
the Internet, to locations with Internet the SCSI remote procedure call NIC
access. It will also allow for smaller model on top of the Internet's
Separate Storage localized SANs to be built using the Transport Control Protocol (TCP).
Assets common Ethernet infrastructure SCSI commands are first divided into
(instead of Fibre Channel or other their various SCSI phases. Then the
Fi
SCSI
mainstream market. Other reasons for encapsulates them into iSCSI
ne
l
this standard's development include: Protocol Data Units (PDUs). These IP Packet
ha n
ne l gaining price point economies of scale PDUs are transmitted over an IP
C Dedicated Storage
Fibr
e
Server
for storage networks by capitalizing on network and reconstructed at the IP Header iSCSI Header SCSI Command/Data
the large volume of Ethernet LAN and IP receiving end. Once reconstituted, the
I
SCSI Storage
S
SC
WAN products, and "easing the use" of SCSI commands are passed to, and iSCSI encapsulates SCSI commands and
SANs by implementing them using these processed by, the target SCSI device SCSI data within IP packets.
well-known Ethernet and IP or system. If required, the
technologies. However, note that using encapsulated SCSI commands and data
Simplified Storage
Storage Client(s) the iSCSI format over Ethernet places a can be encrypted for transmission over
Capacity Management much higher burden on networks than insecure networks.