- Available exclusively for vCenter Server Appliance (vCSA)
- Consist of three nodes – active, passive, and witness nodes
- Passive and Witness nodes are cloned from the existing vCSA (active node)
- vCenter HA cluster can be enabled, disabled, or destroyed at any time
- There is a maintenance mode to prevent planned maintenance from causing an unwanted failover
- Use two types of replication between active and passive nodes
- Native PostgreSQL synchronous replication for the vCenter Server database
- A separated asynchronous file system replication for key data outside the database
- Two vCenter HA deployment workflows
- Basic: all vCenter HA nodes are deployed within the same cluster
- Advanced: the active, passive, and witness nodes are deployed to different clusters
- There is little benefit to using vCenter HA without also providing high availability at the Platform Service Controller layer
- An external Platform Services Controller instance is required when there are multiple vCenter Server instances in an Enhanced Linked Mode configuration.
- Failover can occur when a host failure, or when certain key services fail
- For the initial release of vCenter HA, a recovery time objective (RTO) is about 5 minutes
I have already known about some of these information when testing vCenter HA in my lab. I highlighted the ones I learned from this white paper.
Source: “What’s New in VMware vSphere”" 6.5” technical white paper
No comments:
Post a Comment