How does vSphere High Availability calculate the memory slot size of a virtual machine?

In vSphere HA slot-based admission control, the memory slot size is the sum of the highest VM memory reservation across all VMs plus the memory overhead associated with the largest VM. This combination ensures the slot is large enough to accommodate the worst-case scenario of restarting the most demanding VM on any available host in the cluster.

Subtracting the overhead would underestimate the required slot size because memory overhead is an additional cost on top of the reservation, not something to be removed from it.

Using the smallest VM overhead would produce a slot size insufficient for restarting the largest VM, because the overhead of the largest VM may be significantly greater.

Subtracting the smallest VM overhead is incorrect on two counts - overhead must be added, not subtracted, and the largest VM overhead must be used to guarantee failover capacity.