A: vSphere HA leverages the management network of each ESXi cluster host to monitor host and VM availability. In vSphere 4.x, this reliance solely on the management network meant that an interruption in management network connectivity could create a host isolation situation. Previously, host isolation events were challenging to resolve and often involved VM downtime to restore the cluster to full functionality.

To resolve this single point of failure, vSphere 5.0's vSphere HA has been rewritten to fall back to a secondary connection if its primary management network connection fails. That secondary connection occurs over the storage subsystem and is used only when the management network is lost. The datastores used in this alternative connection are referred to as heartbeat datastores.

By default, vCenter automatically selects two datastores as heartbeat datastores. Those two datastores can also be selected by an administrator.