Resilience
Failover Capacity.
Can you lose a host at peak? Model the cluster, mark how many failures it must tolerate, and read what is left when it happens: the admission-control reserve, the post-failure headroom, and the verdict.
The cluster
Size the cluster and say how many simultaneous host failures it must survive (N+1 is one, N+2 is two).
8 hosts · lose 1
When it happens
Top bar: today, with the HA admission-control reserve hatched on the right. Bottom bar: the shrunken cluster carrying today's load.
CPU549 of 998 GHz used
consumedreserve 125 GHz
after failure63% of survivors
Memory5,734 of 8,192 GB used
consumedreserve 1,024 GB
after failure80% of survivors
N+1: protected. Losing 1 of 8 hosts leaves the memory of the survivors at 80% — the surviving hosts absorb the failed host’s load with headroom to spare.
Keep peak utilisation at or below 79% of the full cluster to stay green after 1 failure (90% of the survivors).Reserve shown is vSphere HA's default cluster resource percentage policy: 1/8 = 13%. Slot policy and dedicated failover hosts behave differently.