woodycollins
Contributor
Contributor

Guest Performance Monitor Question

Many individuals maintaning an ESX (or ESXi) server generally understand some of the performance monitoring aspects that are taken into account when monitoring the health of an ESX server of cluster of hosts.  And for almost everyone I have spoken with there is something new to always learn.  Recently in the company I work for the ESX infrastructure has changed teams and there is more of an interest in monitoring the ESX infrastructure than the previous team running things. Working with our NOC that monitors pretty much everything in our company I have had problems explaining how things work when a machine is virtualized (fairly frustrating) and I was asked if on a typical load of a Windows Server, not using any of the newer performance counters loaded by the tools, is there any performance counters that can be used to troubleshoot %RDY time problems?

So does anyone know if there is?  The best way I could think of would be that you could monitor the amount of queing involved in the operating system. Does this sound right and am I understanding things correctly?  For example, when a VM has either to many processors or there is a high amount of resource contention for the pCPU's,  the processors of the VM's could potentially be put into a "waiting" state for resources to free up for all the required processor's of a VM to allow for scheduling.  And during that time, from an OS inside the VM's perspective, it would look like the processors simply are not handling the request fast enough and que'ing things?

Or is there something else or even a better way to monitor it?

0 Kudos
1 Reply
satya1
Hot Shot
Hot Shot

woodycollins wrote:

Many individuals maintaning an ESX (or ESXi) server generally understand some of the performance monitoring aspects that are taken into account when monitoring the health of an ESX server of cluster of hosts.  And for almost everyone I have spoken with there is something new to always learn.  Recently in the company I work for the ESX infrastructure has changed teams and there is more of an interest in monitoring the ESX infrastructure than the previous team running things. Working with our NOC that monitors pretty much everything in our company I have had problems explaining how things work when a machine is virtualized (fairly frustrating) and I was asked if on a typical load of a Windows Server, not using any of the newer performance counters loaded by the tools, is there any performance counters that can be used to troubleshoot %RDY time problems?

So does anyone know if there is?  The best way I could think of would be that you could monitor the amount of queing involved in the operating system. Does this sound right and am I understanding things correctly?  For example, when a VM has either to many processors or there is a high amount of resource contention for the pCPU's,  the processors of the VM's could potentially be put into a "waiting" state for resources to free up for all the required processor's of a VM to allow for scheduling.  And during that time, from an OS inside the VM's perspective, it would look like the processors simply are not handling the request fast enough and que'ing things?

Or is there something else or even a better way to monitor it?

Hi in single line  generally we monitor IT Infrastructure to respond the problem proactively or for upgradation purpose if any applicaiton not performing since long time.

Yours,

Satya

0 Kudos