With respect to what?
- physical memory utilization
- load
- storage utilization
- io wait time
- theoretical maximum usage based on VM running on each node (with respect to cpu, memory and storage requirements)
- to-be-backupped VMs
- number of HA VMs