VSAN objects are distributed throughout the VSAN cluster, constrained according to user-defined or default storage policies. Objects will be distributed in such a way that quorum can only be maintained on one side of a network partition, should one occur. This means that, although the VSAN cluster itself might split-brain, no VSAN object can do so, and thus data integrity is maintained across all failure scenarios. If necessary, VSAN will take an object offline (resulting in data unavailability) rather than risk a split-brain scenario that could result in corrupt or inconsistent data.

When VSAN objects are created, the various components of that object will be placed in the best possible location in accordance with policy. If multiple stripes are defined by policy, for example, they are guaranteed to be on different platters, but needn't be on the same host(s). If you have specified the fault-tolerance of a single node and two stripes, and there are greater than six nodes in the cluster, every stripe could theoretically be placed on different hosts with no impact to availability in the event of a single host failure. This allows VSAN to be extremely flexible with component placement while maintaining object availability during failure and maintenance scenarios.