VM performance on different hosts

[ariva7]

Hello i have a question regarding the difference of VM performance on different hosts.

Assume we have two identical VMs (8 Cores, 32GB RAM, on the same storage) with identical OS (Windows Server 2008 R2), the same process list, both on ESX vsphere 5.5 the same version and release.

One Host:

Dell PE M620, 2 x Intel Xeon E5-2670 @ 2,6GHz (first generation), 384GB RAM, CPU benchmark score: 12,458

Second Host:

Del PE M630, 2x Intel Xeon E5-2699v3 @ 2,3GHz (third generation), 384GB RAM, CPU benchmark score: 22,740

I assume that the CPU score on benchmark takes into account the cores number.

There are no other VMs running on hosts. What should be the performance of both VM under the same load? I.E. Should the CPU utilization reported by vm's performace be more or less equal?

They both utilizes only 8 cores from all available.

Right now i can observe that the virtual machine working on slower host has 2 times higher CPU utilization, ad trying to find explanation.

[Davoud Teimouri]

VM CPU performance is related to some factors such as physical host CPU.

The first CPU is an older CPU and the newer CPU has newer technology such as faster bus or etc.

But there is some other factors such as host power policy and the server board or memory or others.

Check the host power policy and configure both in high performance mode.

I didn't have this issue on Intel CPUs but I have same experience about AMD CPUs.

I had to add one more core to a machine when it was hosting by AMD CPU instead of Intel CPU.

Even when AMD CPU was newer and my server was newer generation.

[ariva7]

You were right about the power policy. In my case HP hosts were set in "Balanced" profile.

Set it to Max Performace gave expected boost, so the machines have similar performance regardless of the host.