This article explains why the VMAN chart variable for CPU Ready percent is showing a lower value than the esxtop %rdy value.
esxtop is a legacy VMware tool that VM administrators use to monitor the VM environment (Hosts and VM's) utilization statistics in real time.
The following image shows that the esxtop %rdy value for phlprwssqls02 is at 42.58%:
If you check this on the vCenter, since the esx host is hosted on a vCenter and the vCenter is added as the datasource on VMAN, meaning VMAN gets the data of the esx host from the vCenter, it will show the same (or close, since its realtime) value. See the following:
When compared to VMAN by creating a chart that will show the CPU ready percent of the esx host. It will show a lower value. See the following:
This is not an issue. vCenter and esxtop show "%rdy per VM" as summation of "%rdy per vCPU". This makes sense when you consider "%rdy per VM" = number of CPU cycles lost because or Ready state. Generally, "%rdy per vCPU" > 5 % tells admin that he should look at the issue, "%rdy per vCPU" > 10 % is considered to be critical and admin should definitely fix this. "%rdy per VM" is just summation, to show admin that he is losing efficiency a lot (that is, 10 % rdy per vCPU on VM with 8 vCPUs shows 80 % rdy per VM). The disadvantage of this is that %rdy per VM can possibly go over 100 %.
VMAN respects the fact that values in % should never go over 100 %. That is why VMAN shows "CPU Ready % per VM" as average of "CPU Ready % per vCPU". This makes sense as admin should take attention to VMs with CPU Ready % per VM on 5 % (warning) or 10 % (critical).
vCenter and esxtop show "%rdy per VM" as SUM of "%rdy per vCPU".
VMAN shows "%rdy per VM" as AVG of "%rdy per vCPU".
"VMAN %rdy per VM" should equal to "esxtop %rdy per VM" / "number of vCPUs on VM".
"%rdy per vCPU" values should be equal in VMAN and in vCenter.