We need to be able to compare the VM allocated CPU and memory to the VMi CPU and memory.
Specifically in this use case the VMI memory request is the more important one for the memory overcommit use case.

We would like to once enabling the memory overcommit, how many of the VMs are already overcommited, which means they will have a request that is lower than the request by the VM (guest).

For example:
If memory overcommit is set to 200% and a VM is created with a request for 4GB, the VMI will only request 2GB.

But, old running VMs will still have the 1:1 request, until it will have a new VMI.

Once we have these metric we can correctly calculate the actual overcommitment.

Note: This feature require kubevirt VEP and detailed design document.