The following is a summary document that wraps up the Ask The Expert: Performance Calculations on Clariion/VNX discussion:
The way we experience performance is the way people react to various response times. We talked about this a lot since this is the starting point of performance troubleshooting. We’ve discussed a number of tools one can use to check for performance problems: Unisphere Analyzer, Analyzer Helper, SolarWinds and a couple of custom tools. Once you know there is a performance problem, you start to go deeper and try to find out what’s the cause.
Milliseconds, write penalties, I/Os per second… the difference between the various drives, cache and RAID levels. Database requirements, connectivity settings and QFULL errors. And don’t forget the differences between regular LUNs, META LUNs and pool based LUNs. We briefly touched FAST VP and Fast Cache: new technologies that can speed up your system and/or improve TCO. Regarding Fast Cache: in the new Inyo (R32) code data is analyzed in such a way that data isn’t cached on EFD if it’s sequentially written!
For the calculation savvy: what’s the impact of running Snapshots on your storage? What if you’re running remote replication and you loose a link? And which protocol should you use for your links: iSCSI, FC or FCoE?
It was a good interactive session, due in no small part to the TweetChat we held (#EMCATE). Thanks a lot for your questions and posts and we hope to see you again on ECN or Twitter!
Jon & Rob