1

Looking to measure the performance of a VM over a two-week period, which means the Performance tab in vSphere client/UI is not really feasible.

esxitop appears to only track the host, not a particular VM, although not sure if there's a method of filtering on certain VMs.

Solarwinds Virtual Manager has a pretty slick means of doing so, but it's a bit spendy.

So is perfmon accurate enough to measure say IOPS/latency from a storage perspective? I know with the VMTools installed, there are some VMware counters related to memory and cpu that are accurate, but there's nothing for disk counters.

Seems to be very little data about this recently (last 5 years or so) from my searching.

gravyface
  • 13,957
  • 19
  • 68
  • 100

2 Answers2

1

The vendor doesn't want to highlight their own limitations to a client, you'd need somebody on the inside to air their dirty laundry. However, you can infer based on what the Vendor does and what they don't say.

In this case, Microsoft created a whole new set of performance counters in the Hyper-V Hypervisor and Virtual categories. Then you find this documentation from 2017 that specifies where to look for performance bottlenecks.

Only 2 of those categories, Memory and Storage, reference counters inside the virtual machine. So perfmon should be accurate for storage performance.

spacenomyous
  • 1,319
  • 7
  • 15
0

If it's temporary, I'd use a trial of vRealize Operations Manager.
It will provide you the most in-depth look at what's happening.

If you need something quick and dirty, try running a NewRelic server agent to get a view from the OS level.

ewwhite
  • 197,159
  • 92
  • 443
  • 809
  • Yeah, and there's Solarwinds Virtual Manager that has a 30-day trial. Really would like to know if perfmon can be trusted, but nobody wants to touch that (I've asked senior sales engineers and they all push their own tool). – gravyface Oct 31 '17 at 15:17