reading esxtop metrics for analyzing load, Ready Time

[madhavip]

Hi,

I am looking at esxtop metrics to analyze the present load oh the host and if we can add more VMs on it.

Trying to understand the Ready time metrics and what it conveys.

Host config:

Model Proliant DL585 G1

8 CPU X 2.199 MHZ

16 VMS(3 2vcpu vms and 13 1cpu Vms)

RAM 32GB

for few VMs constantly ready time is above 15%, and going up to 30-40% once in a while. when I go for detailed world view, vmm0 and vmm1 processes having the ready time. but %Used is less only, not more than 40% all the time and %CSTP and %MLMTD are 0 and overall CPU Utilization is average 40%. With this I understand that there is no CPU contention at the host level, but something is there with the VM itself.

Also noticed that mostly these are 2 vCPU Vms, though sometimes it is high for 1 vcpu VMs also.

does this indicate that these 2 vCPU VMs need more resources? but %used is not that high.

please help understand, attached the screenshots for reference.

Thanks

M

[DSTAVERT]

Have a look at http://communities.vmware.com/docs/DOC-9279

Forum Upgrade Notice - We will be upgrading VMware Communities systems between 10-12 December 2010. During this time, the system will be placed in READ-ONLY mode.

[MauroBonder]

have too http://www.yellow-bricks.com/esxtop/

[madhavip]

Thanks for the links.

Sorry for late reply. Actually I have gone through those docs earlier, but still wanted to cofirm what the metrics in my scenario(in those snapshots) signifying.  because there is no %CSTP or %MLMTD, but still %readytime is high for 2 vCPU VMs and somtimes for 1vcpu VMs. This doesnt seem to be CPU contention at host level, but is it about the specific VMs. does it require additional vcpu etc?