Specifications

www.ibm.com/support/techdocs 12

Summary of Best Practices for Storage Area Networks

• ISL to switch Edge-1-port_13

• SVC1-node 1-port 10

In SAN environments with redundant fabrics and switches from the same vendor are

deployed, use the same topology in the redundant fabrics. Using the same topology will

be easier for administrators to keep the workload balanced between fabrics. Some

exceptions are to be expected, such as single data paths to certain tape drives.

Overall, the number of paths between a given server and associated disk storage

resource should be balanced between redundant fabrics. Further, one of the redundant

fabrics should be capable of handling the total traffic workload for the SAN environment.

If the SAN environment is composed of more than two redundant fabrics, the complete

failure of one fabric should not overwhelm the other fabrics. Consult Section 6 on High

Availability topics for more details on fail-over scenarios.

Locality is the term used to describe the aspect of a SAN design where devices with

high traffic levels between each other are connected and zoned across the same

switch. One example is the nodes of an SVC cluster on the same switch as the back-

end disk systems utilized by the SVC cluster. Thus, high degree of locality means less

traffic over ISLs between multiple switches and thus less risk for latency and/or fabric

congestion. The implication with locality is that the SAN administrator must have a clear

understanding of all traffic patterns to determine which SAN-attached devices should be

collocated on the same switch. In certain situations, the situation is very clear, such as

a TSM server and its associated tape drives; but not always. Either way, a high degree

of locality within a fabric is preferred.

As newer devices are incorporated into a SAN environment, there is a tendency for the

newer devices to utilize newer hardware which are capable of higher port speeds than

existing fabric devices. As these newer devices are added to the existing SAN

environment, the attach point of these devices needs to be considered. Simply stated,

devices capable of higher port speeds which will be heavily utilized by existing SAN

devices should be located closer to a fabric’s center. The reasoning of this approach is

a result of the fabric center (the core in a core-edge design) has sufficient connections

and bandwidth radiating outwards to the edge for the additional workload.

It is a common practice for storage systems to share a given port among multiple HBA

ports on multiple servers to increase the level of connectivity to the storage system.

The server-to-storage port ratio is called the fan-out ratio. As the size of fabric grows,

the number of switches in a given fabric is very likely to correspondingly increase to the

point where servers are connected to one fabric switch and their shared storage port

connects to a different switch. Thus, the server-to-storage traffic must traverse at least