3

We are monitoring several Windows Server 2012 R2 VMs running on Hyper-V. Now SCOM (2012 R2 UR2) reports:

The threshold for the Logical Disk\Avg. Disk sec/Transfer performance counter has been exceeded.

This alert is raised if the counter value "Avg. Disk sec/Transfer" is greater than 0.04 on 15 samples. This corresponds in our case to a duration of 15 minutes.

I am suspecting this to be a false alert. There has been a bug in the past where calling the function RegQueryValueEx() on W2008 SCOM-clients returned false data (http://support.microsoft.com/kb/2495300/en-us).

Now I am wondering if there are any similar issues or known bugs that apply to W2012 R2? Google did not help here. Any input or experience on this? Thanks in advance.

[Addendum]

After double checking the values it turns out that they are getting correctly reported.

Big question: What is the best-practice or recommended threshold value for a virtual machine regarding this counter? Microsoft recommendes this value not to exceed 0.04. Is this applicable to VM as well?

Matthias Güntert
  • 2,358
  • 11
  • 38
  • 58

1 Answers1

4

This is not a false alert. It shows disc latency in excess of 40ms. Whether or not that is a PROBLEM is your definition - and you should adjust things. Acceptable disc latency really depends on whether the systems get too slow and it is a broad topic and one that has to be defined by you.

On secondary VMs I accept 50 or even 100ms at times, on my main database servers 5ms is a problem (hence they run on SSD).

TomTom
  • 50,857
  • 7
  • 52
  • 134