Specifications

www.ibm.com/support/techdocs 40

Summary of Best Practices for Storage Area Networks

general guidelines or suggested threshold levels to determine when a connection and/or

device is performing well or poorly. Since each network is unique, meaningful

thresholds will need to be determined via a trial and error method.

One alternate method involves using TPC to collect performance metrics over a period

of time, such as a full week, and then use the various peaks as a starting point for

setting thresholds. A key proviso with this method is that SAN operations during the

initial monitoring period must be normal without any notable disruptions or abnormal

activities. Needless to say, performance metrics should be collected on a regular basis

(such as once a month and no longer than quarterly) to better monitor changes in the

SAN environment over time.

If necessary, thresholds may warrant modifications. However, a threshold should never

be set much higher than 80% of a connection’s bandwidth. Levels over this threshold

level are an indication can either additional port-channel bandwidth are needed or that

the workload on a particular device port may need to be reallocated to another port on

the same device, or moving some of the workload to another device.

9.1 Performance and Metrics

The answer to some performance questions is the same two words; it depends. This

is not much use to you when trying to solve storage performance problems, or rather

perceived storage performance problems. But there are no absolutes with

performance so it is truly difficult to supply a simple answer for the question: “what is

a good performance number for a VDisk?”

The best rule of thumb metrics for any system are derived from current and historical

data taken from specific configurations and workloads that are meeting application

and end user requirements. Collect new sets of metrics after configuration changes

are made to important storage resources. Keep a historical record of performance

metrics.

It can be better to move users up the performance spectrum rather than down.

Users will rarely complain if performance increases. So, if there is uncertainty as to

the correct pool use the one with lower performance, and move them up to the

higher performing pool later if required.