In addition to network heartbeats, storage heartbeats were introduced to increase the resilience of the infrastructure.

By default, two different shared datastores will be used for storage heartbeats, again to increase the resilience of the infrastructure. It is possible, however, to configure an advanced setting, das.heartbeatDsPerHost, to change the value to a different number of datastores, or das.ignoreInsufficientHbDatastore, set to true to ignore the warning.

The following screenshot is the configuration page for Heartbeat Datastores:

Storage heartbeats are used only when network heartbeats have failed. Each host will create a specific file on the shared datastore that can be accessed by all other ESXi hosts. This file, host-XXXX-hb, is an empty file. The corresponding ESXi hypervisor exclusively opens that. If the ESXi hypervisor loses access to the storage, the master ESXi node will discover that the lock from the file disappears, meaning that the host has lost access to the storage and an HA event will occur.