Proactive HA is a new feature integrated with server vendor monitoring systems. The idea is to provide HA to VMs even before an actual failure occurs. For example, the server has two redundant power supplies, and one of them fails. There is no direct impact on the ESXi server itself, but this condition might lead to downtime when the second power supplies fail. With proactive HA, you can evacuate the VMs from the affected ESXi hypervisor event before it becomes unresponsive.

When any of the hardware fails, it is marked by the hardware monitoring agent as unhealthy. Based on the configuration, vCenter Server will classify the hardware failure as degraded or severely degraded and the host is placed in a specific state called quarantine mode.

In quarantine mode, DRS won't use the affected ESXi server for the placement of any new VMs, and it might even proactively evacuate the VMs of the ESXi host. Evacuation will only take place in situations when the migration will not affect the performance of the VMs. There is an option to place the host into maintenance mode as the result of degraded state.

Remediation actions available with proactive HA include:

• Quarantine mode for all failures: Quarantined hosts are not used to run new VMs, but currently running VMs will still be run on the top of the quarantined host.
• Mixed mode (quarantine mode for moderate and maintenance mode for severe failure): If the host suffers from moderate degradation, the VMs will be kept running on the host, but new VMs will not be run on the host. However, all VMs will be migrated off the host in severe failures.
• Maintenance mode for all failures: All VMs will be migrated off the host no matter whether moderate or severe degradation occurred.