In a VMware high availability setup, how many hosts are typically required to ensure redundancy?

In a VMware high availability (HA) setup, having three hosts is generally recommended to ensure adequate redundancy. A three-host configuration allows one host to fail while still maintaining the ability to run the workloads on the remaining two hosts. This configuration supports the HA feature that ensures virtual machines (VMs) can be restarted on other available hosts in the cluster without a loss of service.

The rationale for having three hosts is based on the need to maintain a quorum for cluster operations. In the event of a host failure, the remaining hosts can still sustain operation without compromising the overall cluster availability. Additionally, having an odd number of hosts helps to avoid split-brain scenarios in which two groups of hosts could potentially take control of the same resources or VMs.

In practice, while two hosts can provide some level of HA, they do not deliver the same level of reliability and failover capability due to the lack of redundancy in managing network partitions or failures. A four-host setup may provide additional capacity for workloads but does not significantly enhance redundancy compared to having three hosts. Therefore, the three-host configuration strikes the best balance between redundancy, resource management, and simplicity in high availability environments.