I'm having a bit of trouble understanding vROps and ESXTOP. We have a database VM that is currently configured with 30 vCPU's and 96GB of RAM. Last week the application owner was stating they are seeing sluggish performance on the SQL database VM and per his findings via SQL queries there are high CPU wait times.
I've looked into vRealize Operations Manager and it suggested we go from 30 vCPU's to 41 vCPU's:
I also SSH into the host the VM is on and looked at ESXTOP and the VM shows to have what I believe would be considered high ready time (30%), see highlighted line:
I'm not understanding how vROps is suggesting to adding so many vCPU's.
Any help on how to better understand this issue and vROp suggestion?
Let's first get an understanding of your "stress" policy settings, which is feeding your first screenshot. Look at what I'm attaching 'here' and send us a screenshot of your policy that is currently getting applied to the VM in question.
The default stress metrics are used.
Assuming default settings, Stress is reflecting CPU demand and Mem demand. If your CPU Ready is 30%, you've got host congestion problems. I would not take the CPU resizing recommendations honestly until you get that contention under control. If you correct the ready % problem, I expect you're performance will increase and your CPU deficit will be reflected more accurately.