Performance chart inconsistent

[NuggetGTR]

Hi All,

Ive come accross an issue recently with the performance charts in virtual center and was wondering if anyone has seen this and know how to fix this?

I was looking at the cpu usage and cpu ready time on a couple of virtuals after a ticket came in about poor performance. When looking at the real time chart cpu ready was about 100 ms with one spike in the hour to just under 1000ms. when looking back to the past day chart it was showing up to 1000000ms cpu ready.

was thinking thats a bit odd and continued to monitor the real time for the majority of the day knowing it was sitting on about 100ms for most the day. I looked at the past day and for the few hours i had been monitoring it,, it was showing about 5000ms where i know it was at around 100ms for the majority of it. I ran esxtop and it seams to be on par with the realtime chart. i checked a hand full of other virtuals and looks to be suffering the same inconsistencies.

any ideas as to why this is?

running esx 4 u1 for all my hosts and vsphere vc.

Cheers

[AntonVZhbankov]

This is because it shows you sum of CPU ready during minimal time slice on the chart.

---

MCSA, MCTS Hyper-V, VCP 3/4, VMware vExpert '2009

http://blog.vadmin.ru

[NuggetGTR]

Sorry dont quiet understand, you saying the cpu ready chart of the past day is showing me the sum total of the cpu ready for a slice of time at which it marks it on the graph?

but real time is showing me what it "actually" is at 20sec int

and how big is the default time slice for the past day graph? sorry dont have vc infront of me atm other wise i would have a look to see if it was there.

[mittim12 · Accepted solution]

The past day is a roll up of 5 minutes. Check out this link for some good information regarding performance monitoring, http://communities.vmware.com/docs/DOC-5230.

If you found this or any other post helpful please consider the use of the Helpful/Correct buttons to award points

[NuggetGTR]

awesome thanks, never read that before, To me it seams a little odd to to have the sum of it for that 5min slice would have through average would better show the performance history at a glance without having to bust out the calc

[AntonVZhbankov]

I absolutely agree with you, but we have what we have.

---

MCSA, MCTS Hyper-V, VCP 3/4, VMware vExpert '2009

http://blog.vadmin.ru