I am curious if anyone is using the new side-channel-aware scheduler and what your experience has been. I ran the analysis script and everything was green (No expected impact...). While I didn't notice a real increase in CPU utilization, I did notice an increase in CPU Ready Time and Co-Stop Time. I am not sure those values were considered by the readiness script. What has your experience been? I never had an issue with Co-Stop, now I have some VMs hitting 3+% fairly often and Ready Times have doubled.
I have applied it to my hosts currently there does not seem to be any detrimental affect, that said, our hosts are currently not that well utlized (green field cloud - growing tenants) we are monitoring to see just how much of a performance hit we get when we ramp up density.
we are considering running a VMmark tile run against a cluster that has not to date been remediated and then against one that has (all our hosts are exactly the same).
Thanks for the reply. Here is a graph of CPU ready time on 1 host. It is a summation on a 20-core server and it is a good representation of what I am seeing on most of my hosts after enabling the scheduler. The almost flat line is how the host looked for days/weeks leading up to the change, the break is the reboot, and the spike is the result.
Any update to this?