dmanrez
Contributor
Contributor

How to troubleshoot once you have reviewed the graph

Hi

I'm looking at the graph below and noticed that the spikes are high. I was wondering if this is normal for a Host Server? How do I go about troubleshooting the issue to reduce the spiking.

0 Kudos
5 Replies
oreeh
Immortal
Immortal

FYI: this thread has been moved to the Performance & VMmark forum.

Oliver Reeh[/i]

[VMware Communities User Moderator|http://communities.vmware.com/docs/DOC-2444][/i]

0 Kudos
weinstein5
Immortal
Immortal

This vert well could be normal depending on the load the ESX server is handlings - the more ciritical metric to keep an eye on is Ready Time - amount of time a VM is waiting to be schedule - this is the best indicator if the VMs are being contrained by not enough CPU resources -

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful
0 Kudos
FredPeterson
Expert
Expert

You have to remember the goal of virtualization is consolidation of hardware resources in order to maximize what would normally be minimized. Providing more information such as number of guests on this host and the number of guests that have more then one vCPU would help as well.

Looking at the host itself here will only give you an idea of the host, not the guests. Also take note that the largest peaks are MHz used. Get rid of the MHz data points and look just at the Cores themselves. You'll take note that your average CPU otherwise is < 50%

You need to couple this graph with the same graph of the guest CPU usage to determine if only a few select guests are using a lot of CPU or if its just wide spread sporadic.

Then I would look at CPU Ready Times to determine if this is even anything to be concerned about. High utilization and low ready times == excellent consolidation with maximum resource usage and minimum guest server impact. But if this high utilization is also coupled with high ready times...you've got an issue to worry about.

0 Kudos
V_Ahillen
Contributor
Contributor

What is considered an acceptable ready time? Is there a reference guide somewhere? I am interested to see what is considered good and bad for 1, and 2vCPU guests.

0 Kudos
weinstein5
Immortal
Immortal

It will depend - CPU depndent applications will have a much lower tolerance for a Ready Time that might be acceptable to an application that is not as dependent on CPU - you might check this out http://communities.vmware.com/docs/DOC-10352

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful