While VMware HA was initially introduced, one key aspect when designing the HA cluster layout was to ensure that hosts are distributed across racks/chassis to avoid a single point of failure in terms of racks or chassis. This implementation has trickled up into the VSAN layer. In its first generation, objects were distributed across hosts in a cluster and the number of failures to tolerate ensured that the host failures could be sustained. However, this did not factor into account whether the hosts were part of the same rack or another rack. Hence, there could have been a situation wherein all the hosts holding the object may end up being on one single rack. VSAN 6.0 introduces an enhancement with fault domains that have rack awareness. This effectively mitigates a single point of failure at the rack level.

In the following figure, we see a three-node cluster stacked in the same rack and connected to the same top of the rack switch and power outlet. This entire stack can...