We have a cluster where vROPs is reporting CPU demand (excluding overhead) to be about 40% higher than the CPU usage. Generally this would point towards CPU crunch on the ESXi host however all the VMs running on the cluster have CPU ready and co-stop values within threshold.
Worst CPU ready is less than .5% and worst CPU co-stop is less than 2%.
vRops also recommends to downsize some VMs and no VM has recommendation to increase vCPU - this would mean VMs have more than what is needed so overall demand should be at par with what VMs have .
BIOS power management is set to High performance as well.
What else can be checked to identify this gap in values reported for CPU usage vs Demand.
Can oversized VMs cause this although it does sound counter intuitive.