Hi all,
I'm managing some SAP HANA (monster VMs) and I'm experiencing quite high CPU ready values (sometime esxtop shows 30% %rdy values), but without visible performance issue.
For example we have a quad socket server (24 core x 4 socket x 2 thread / 1.5 TB RAM) that host only one HANA VM (96 vcpu 1.1 TB RAM).
Is there anyone who can share a similar experience with monster VMs?
If someone else has experiences with VM that exceed 1 numa core of the host, could share any suggestion.
I'm not able to understand if those values could hide an issue or not.
Thanks a lot for any help.
Bye
If the screenshot is showing where you look then the explanation is that esxtop shows a sum of the per vCPU metrics for the sched group (VM), so it actually isn't a lot per vCPU.
I'm talking about "Ready" for ~40 minutes here: https://www.vmworld.com/en/video-library/video-landing.html?sessionid=1589484575728001Zb6J (~24 min in) if you want to know (most of) what there is to know about that particular metric and how we track / display it.
If the screenshot is showing where you look then the explanation is that esxtop shows a sum of the per vCPU metrics for the sched group (VM), so it actually isn't a lot per vCPU.
I'm talking about "Ready" for ~40 minutes here: https://www.vmworld.com/en/video-library/video-landing.html?sessionid=1589484575728001Zb6J (~24 min in) if you want to know (most of) what there is to know about that particular metric and how we track / display it.
Thanks a lot, very interesting video.
Surely this helps me to understand ready and readiness correctly.
Have a nice day.