0

Are there any tools on Windows that allow performance statistics to be recorded? Something similar to iostat 10 > iostat.log on Unix?

I have 10 Windows VMs that run a whole lot of regression scripts overnight. Sometimes we see failures that we suspect are caused by over-high disk usage. But at the moment I have no way to tell. What I'd like is a log file in the morning containing statistics for the night (eg CPU, memory, disk I/O etc), so that I can see if failures correspond to eg high disk I/O.

I don't know if it makes more sense to get stats per-VM, or from the underlying hardware. Any advice appreciated!

2 Answers2

1

check the performance monitor (sorry, I only have a Dutch xp handy, so the translation could be pretty wrong). You can add counters for the physical or logical disks for about everything you need about the disks performance. You can script it all with logman.exe.

I use the windows perf monitor a lot in combination with the check counter of the check_nt plugin for nagios to get historical data.

natxo asenjo
  • 5,739
  • 2
  • 26
  • 27
  • Perfmon is the correct name. @Sam: You should also consider using a tool like SQLIO to benchmark your VMs. The name is a liitle misleading, it really has nothing to do with SQL. – Robert Kaucher Oct 07 '10 at 17:42
0

Get stats from the hypervisor - I rarely trust 'in-VM' stats from OSs due to timeslicing errors - plus you'll probably get more useable data - you don't mention which hypervisor you're using?

Chopper3
  • 101,299
  • 9
  • 108
  • 239