SAN performance can be a far-reaching topic. It can cover everything from basic storage array performance (things like IOPs and throughput) to I/O latency (as in how many switch hops to storage) to WAN bandwidth and latency for remote data movement.
Storage array MIBs can provide some information and network components may be involved within a SWAN (storage wide area network). I find the best method is to view things from an application perspective. You can use tools like NT's PERFMON to gather statistics on how the storage network is working by how the host sees it. To weed out fabric-based bottlenecks you can use the actual fabric switches to get performance metrics on a per-port or per-subsystem basis.
Use a tool or a combination of tools that can provide this information. Start at the host, work into the fabric and then into the storage subsystem itself. After all, it always comes down to the actual spinning disk for storage performance in the long run. Most storage vendors provide a graphical performance tool with their subsystems to get good historical or real-time stats.
If NAS is involved, caching can help a great deal as it does in subsystem performance. For NAS though, also look at the IP Network bandwidth, IP stack efficiency, CPU utilization on client host and NAS array and congestion on the network.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in our Storage Networking discussion forum.
This was first published in February 2002