We have been getting very poor performance from CA APM.
It takes 5 minutes or more to see statistics.
CPU usage is %1 and there is a lot of memory.
Disk IO is almost negligible.
I found a paper and followed the guidelines there (performance tuning paper).
What else can we do?
I am not sure which link that you looked at but here are the ones I generally provide:
-- Application cluster performance check.. Start here
CA Tuesday Tip: Top 10 common causes for EM performance issues - Checklist
CA Tuesday Tip: APM Health Check Tips Good overview
CA Tuesday Tip:MARCH 2013 - Top 20 common causes for EM performance issues
Looking at your perflog and your logs may give clues. Your architecture including what's in your SmarStor, cluster settings/setup may be all contributing to this. There may be multiple things needing to be fixed here. Once stable, an ongoing healthcheck will keep things running smoothly.
I suggest opening up a support case to help dig deeper into what is going on and take care of some of the "quick hits."
Meanwhile you go through the links and suggestions provided by Hal, I want to share my experience where I had similar issue of late responding APM.
So could you please let us know something about your architecture and deployment of CA APM like how many clusters, how many agents, metric count etcBecause over-crowding of agents resulted for me this kind of issue. [including clamping for me]. Bdw which version are you using ?
Also could you please check the network health between your Wily EM\MOM and the client.
Last but not the least, if this is the case with thick client only, try to increase the heap assign to workstation in IntroscopeWorkstationproperties file and share the difference.
Hope it helps.
Please let us know if additional assistance is needed or what you was provided is sufficient