Enthusiast
Enthusiast

Which VM Memory Metric?

Which memory metric should I be using to measure a VMs memory use? An end user recently asked me why the memory usage inside of their VM (i.e. Task Manager) is different from what they see in their vROPS dashboard (both in real time and historical). A legitimate question that I've never asked myself.

To measure a VMs memory use in vROPS, assuming it should mirror what the OS is reporting, which metric should I be using? Guest Active Memory (%), Usage %, Workload %. something else?

pastedImage_0.png

0 Kudos
5 Replies
Enthusiast
Enthusiast

Thank you daphnissov. Heres my takeaway.

  1. Memory utilization in the guest OS (Task Manger) will almost always be greater than "Active Memory". This is problematic because I dont want users to think (when viewing vROPS) that their machine is less busy than it actually is.
  2. Memory utilization in the guest OS (Task Manger) will almost always be less than "Consumed Memory". This is problematic because I dont want users (when viewing vROPS) to think their machine is busier than it actually is.
  3. The metric "Memory|Usage (%)" is an aggressive percentage.
  4. The metric "Memory|Workload (%)" is a conservative percentage.
  5. The metrics "Memory|Utilization (KB)" and "Guest|Needed Memory (KB)" are nearly identical. They are used by the capacity engine and rightsizing recommendations.

After some testing (see screenshot below), it seems the  "Memory|Usage (%)"  and "Memory|Workload (%)" metrics are the closest to what the guest OS (Task Manger) reports. "Active Memory" and "Consumed Memory" are too extreme compared to Task Manager.

Based on the testing (and the author's recommendation in the comments section of the article you provided), I think I should use the "Memory|Workload (%)"  metric in my VM dashboard going forward. Would you agree that is the best solution to give users an accurate representation of their VMs memory activity?

The "Memory|Utilization (KB)" and "Guest|Needed Memory (KB)" metrics seem to be useful as well, especially as the are used by the capacity engine and rightsizing recommendations. I will be including these metrics as supplements in my VM dashboard.

pastedImage_15.png

pastedImage_1.png

0 Kudos
Expert
Expert

Yes @nmanm0305  you are rite in understand it correctly.

As daphnissov  provided you with rite doc , based on many user issues with Memory metrics , Memory|usage should be the best one to rely on as if now.

Please mark this as "correct" or " Helpfull" if this answers your query.

Regards

Gayathri

0 Kudos
Expert
Expert

Thanks. but in that article suggest to use memory utilization . Is that correct ?

Please mark helpful or correct if my answer resolved your issue.
0 Kudos
Commander
Commander

I just tell my end users not to use the memory metrics in vrops as it is a mine field trying to explain the differences to application users that just about know there own name to log in in the mornings.

Made easy for me as we "officially" dont use vrops for VM monitoring