HPE Storage Users Group

Hi,

I wondered if I could get some assistance getting a baseline of our 3Par performance. Basically I would like to collect some stats on how things are looking at the moment so that I can compare it against previous months and also going forward.

We have 80 disks of which 56 are FC 10k and 24 are NL 7k. I worked out that we have a total of 7840 IOPS based on 140 IOPS for the 10k disks and 1440 IOPS on the the 60 IOPS 7k disks. I have been running reports from the SSMC using the Physical Drive Performance reports to try and get an average IOPS usage but I am unsure if this is an accurate way of doing things. How would I be best collecting this and any other useful information to help with troubleshooting any performance issues?

Thanks,

Graham

Assume 0 iops for 7k if you are trying to size an upgrade, don't buy 7k assuming they'll give you performance, just capacity.

Take a look at this post, covers back end, controller nodes and front end.

download/file.php?id=261

It's important to keep and eye on the backend PD's so you can assess easily when things are reaching saturation point. Changes to read write ratios and IO size can have a big impact on back end performance while IOps may remain negligible on the front end although first sign is usually a front end increase in latency.

If you had system reporter, this would be easier for me to answer. Our setup runs system reporter and logs/keeps data for 3 years... so its really easy for us to watch our system load over time and notice major changes.

I need to do some work on this end to catch up to the rest of the world and start using SSMC, System reporter has been a great tool for us, for example.. it took me less than a minute to pull this "baseline of FC physical disk IOPS":

_________________
Richard Siemers
The views and opinions expressed are my own and do not necessarily reflect those of my employer.

Thanks for all of the advice so far, any other tips would be appreciated. I have just installed the new 2.1 SSMC which has the Threshold Alerts which seems pretty useful.