The Business Case for Storage Networks
Increased port capacity and redundancy are popular features of Fibre Channel
SANs. The elimination of a single point of failure in the host-storage relationship
is signiﬁcant and increases business availability and uptime. Figure 2-1 illustrates
the concept of a single point of failure on an application in a DAS environment.
Figure 2-1 Sample DAS Conﬁguration
In Figure 2-1, two non-clustered hosts share a single storage frame without
the beneﬁt of a SAN. If the application residing on Host 1 is not mirrored on Host
2, and the link between Host 1 and the storage array is dropped, then the
application cannot access its storage. There is no way for the disk devices attached
to Host 1 to be shared with Host 2; therefore, the application on Host 1 is
prevented from starting on Host 2. Similarly, if Host 1 goes down, there is no way
to bring up the application from Host 1 on Host 2. This many-to-one conﬁguration
does not offer high availability or data resiliency.
Figure 2-2 shows the elimination of the single point of failure in the same
environment with the installation of two 16-port Fibre Channel fabric switches
and the creation of a SAN.
Chapter 2: The Business Impact of Storage Networking Technology
Figure 2-2 SAN Conﬁguration Examples
In Example 1 of Figure 2-2, both Host 1 and Host 2 have multiple paths to the
disk through two ﬁxed 16-port Fibre Channel switches. The multiple paths
through the Fibre Channel switches create redundancy. In the event of a loss of one
of the Fibre Channel links, the applications on Host 1 and Host 2 can still access
storage over the SAN. Additionally, this conﬁguration supports what is called a
poor man’s cluster. In other words, if the applications are mirrored (via content
management software or via a manual ﬁle synchronization process) across both
Host 1 and Host 2, and one of the hosts goes ofﬂine, the remaining host can (using
volume management software) import the disk groups normally reserved for the
other host and successfully bring the application back online. There is still the
issue of the storage unit as a single point of failure, an issue that can be addressed
with multiple, redundant storage units and more frequent backups.
In Example 2 of Figure 2-2, all single points of failure have been eliminated.
If clustering is enabled at the host and application level, both hosts should be able
to access storage as long as the appropriate volume and logical unit number (LUN)
management software is enabled and properly conﬁgured. This any-to-any
relationship is one of the key beneﬁts of implementing SAN technology.
The ﬁnancial impact of mission-critical application downtime cannot be
overstated. Based on data from the U.S. Department of Commerce Census
Bureau, recent estimates of e-commerce retail sales for the second quarter of 2004
SAN Configuration Example 1 SAN Configuration Example 2