Granular stats

Question

I am currently testing SolarWinds as a replacement for our current monitoring solution. One thing I am still having an issue with is how granular we can get on CPU, Memory and Disk space, specifically in terms of I/O rates (idle, non-idle, time to write etc). We have a blend of many flavors of unix and various windows distros. Anyone have a template that is capable of getting extremely granular that they would care to share?

mesverrum · Answer

Ah i see what you are asking, you want different metrics than what are included out of the box.  I've seen some pretty decent linux monitoring scripts at client sites but don't have any handy.  You might take a look at the AIX templates on the content exchange to see some fairly good examples of how to process commands like sar or dstat into a format that solarwinds will accept as a component monitor.

troyfred · Answer

I think you may have misunderstood what I am looking for. With the current templates applied I get CPU, Memory and Diskspace data, but I mean granular as in average I/O transaction time averages per disk.

mesverrum · Answer

How granular you can get is kind of a function of how quickly the endpoints can respond to the polling requests.  Default SAM templates are all set to collect one datapoint every 5 minutes, Node stats like cpu and mem default to 10 minutes but I've seen many people reduce that interval.  I've not had any client who felt it was necessary to take it down to 1 minute or below but you can stress test it in your environment to see what results you get with your hardware.