Hi, I've read some guides which say to keep the Ready time below 5%.
Should this be per vCPU or for the whole VM?
So if I have a VM which is showing 1000ms realtime I would do
1000 / 20000 = 5%
so do I need to make sure the Ready time is below 1000ms realtime, or do I need to do more calculations because the VM has 4 vCPUs assigned to it?
Thanks
What are you looking at exactly? ESXTOP, vCenter UI?
Hi thanks, vCenter UI mainly
Although I use ESXTOP too
For esxtop the value is in Percentage, and it shows the number per VM by default when not expanded, meaning that you will need to divide by the number of vCPUs you have. So if you see 10 and you have 4 vCPUs then the number would be ~ 2.5% per vCPU.
When I look at vCenter Server in my lab I can look at two metrics when I select advanced monitoring:
Readiness: average in percent
Ready: summation in ms
It shows it for the VM, as well as for each vCPU
Thanks for this!
In a lot of articles it mentions 5% as being the number to not go over
is that 5% per VM
or 5% per vCPU?
thanks
that is per vCPU, and I would probably use 10 as a threshold myself.
What Duncan said.
With regards to the threshold, it depends a lot whether that is peak, sustained, avg. over a certain interval etc. 10 is certainly a good alerting threshold for further investigation, ultimately the impact to your applications and their requirements would dictate the acceptable threshold. For those applications that can't deal with an alerting threshold that weeds out minor and intermittent contention, set reservations.
If you want to learn more about Ready Time, check out: https://www.vmworld.com/en/video-library/video-landing.html?sessionid=1589484575728001Zb6J (~24th minute).
P.S.
VMware with a lower case w please 🙂