Virtualization Host Sizing Calculator

Host sizing starts from the workload: the vCPUs to place are the VM count multiplied by the vCPUs per VM, and dividing that total by the target vCPU-per-core ratio gives the physical cores the scheduler needs, which is then spread over whole hosts. Memory is counted separately as VM count multiplied by RAM per VM against the usable RAM on each host, and the larger of the two host counts decides the cluster — that is the binding constraint. Spare hosts for HA carry no workload, so they are added after the base count and raise memory capacity and core headroom without lowering the achieved ratio. The achieved vCPU-per-core figure is the total vCPUs divided by every core in the sized cluster, including spare hosts, so it lands at or below the target after rounding up. Memory committed compares guest RAM with the memory of all hosts including spares. Cores are physical cores rather than hyperthreads, usable RAM excludes the hypervisor reservation, and the model assumes a uniform VM profile and no CPU reservations or limits: it is a first-pass planning estimate, not a vendor sizing certification, and per-VM demand, NUMA layout, storage and network bottlenecks, licensing and failure-domain placement are not modelled. Everything runs in your browser with no storage, uploads or network access.