VM Optimization problems

[matai]

I've been running VM Optimization reports with no problem. What I'm running into is if I change the specs of a VM to what the report recommends, we receive reports of slow VMs. The VMs running Windows show the CPU and RAM pegged to the max.

There seems to be a big difference between the Veeam One report and what Windows is actually using.

Does anybody have any suggestions on how to best right-size VMs with Veeam One?

Thanks

[Shestakov]

Hello,
Could you clarify what exact reports you are talking about?
For VMs` sizing we recommend Oversized VMs and Undersized VMs reports. These reports are primary generally based on CPU Usage, Memory Active, Memory Consumed etc. metrics taken from vSphere.
Also Idle VMs and Powered Off VMs reports help to reclaim resources.
Thanks!

[Vitaliy S.]

...and a couple of other links for additional details > undersized and oversized vm reports and Differences between Task Manager and Veeam

[matai]

Hello,
Could you clarify what exact reports you are talking about?
For VMs` sizing we recommend Oversized VMs and Undersized VMs reports. These reports are generally based on CPU Usage, Memory Active, Memory Consumed etc. metrics taken from vSphere.
Also Idle VMs and Powered Off VMs reports help to reclaim resources.
Thanks!

I've been using the Oversized VMs report. We have zero undersized VMs.

[matai]

...and a couple of other links for additional details > undersized and oversized vm reports and Differences between Task Manager and Veeam

I understand how the guest versus host reporting is different and that makes sense to me. The problem is that I'm the only one on my team that seems to get that. I reduced resources on some pools and standalone VMs to match what the Veeam ONE Oversized VMs report suggested. Now we are having users complain that their VMs are slow. Of course, that can't be quantified, although one of the other engineers argued that it did take longer to open programs and run some tasks.

So if I right-sized a VM according to what Veeam ONE suggests and the user experience degrades. What course of action do you guys suggest?

Thanks!

[Shestakov]

Well, I would ask for CPU/Memory statistics from the engineer to find out if the load is critical.
If peak usage is around 95-100% and average is pretty high as well, it makes sense to add some resources to the VM.

A trade-off between performance and resource allocation varies from user to user, and depends on their demands and capabilities. So it`s up to you to decide.

[Vitaliy S.]

BTW, did you use memory active or memory consumed counter in the over-sized VMs report?

[matai]

BTW, did you use memory active or memory consumed counter in the over-sized VMs report?

I was using active, should I use consumed?

Thanks

[Vitaliy S.]

For most of the VMs memory consumes would show how much RAM is used by the VM. On the other hand, for applications such as SQL Server/Exchange this counter will always show 100% of consumed memory, no matter how much RAM you allocate to these VMs.

Please take a look at the existing discussion about these metrics > Oversized VM's Report Too Agressive

Hope this helps!