VMware Cloud Community
jkopp
Enthusiast
Enthusiast

DRS - Active vs Consumed CPU usage

We've recently upgraded our ESX farm from 3.5Update1 to ESX4 build 175625. After doing so, we've experieced a 60% drop in host CPU utilization (what used to consume over 40% on 4 hosts now cosumes less than 30% on three. While this is great news for us, I'm afraid it's too good to be true. One thing I have noticed is that my DRS stats are screwy. While I don't have a CPU bottleneck (over comittment of resources), the DRS Distribution chart looks as though something is wrong. See attachment for clarification. The mail gateway in question is running the latest ESX hardware and VM Tools.

What's the difference between consumed/active/DRS CPU Mhz and why aren't they all the same and running at 100% when my host utilization is at less than 25%.

An open ticket with VMWare has yet to explain what's going on.

Thanks for the help! Jason

0 Kudos
3 Replies
Yattong
Expert
Expert

Hey,

Consumed - shows the actual consumption of resource by the virtual machine

Active - shows the estimated amount of resources consumed by the virtual machine if there is no resource contention.

Which makes sense, being '29% resource only delivered' in the graph.

What doesnt make sense is that your seeing under utilization on your hosts. You may need to double check all your share/res/limit settings if you have any.



If you found this or any other answer useful please consider the use of the Helpful or correct buttons to award points

~y

If you found this or any other answer useful please consider the use of the Helpful or correct buttons to award points ~y
jkopp
Enthusiast
Enthusiast

That's my point exactly. I don't know what I have misconfigured at this point. I don't have any resources pools set at this point and just let the VMs take what they need from the host.

On the guests there is no reservations or limits and CPU shares set to "normal". I do have the "shares" set to "low" on mailgw2, but with all this idle CPU, I wouldn't think that would matter.

0 Kudos
jkopp
Enthusiast
Enthusiast

I believe this is due to CPU contention on the hosts. Each host had a maximum of 8 cores. I'm finding out that some of our VMs which are SMP don't get addequate CPU time and were suffering from "CPU Ready".

0 Kudos