Failover
Automatically switching to a standby replica when the primary fails, so service continues with minimal interruption.
Failover is switching traffic from a failed component to a healthy standby — promoting a replica to primary, rerouting to another node — so the system keeps serving. It can be automatic (triggered by health checks) or manual.
Worked example: a database primary crashes; the cluster detects missed heartbeats, promotes a replica to primary, and points writes at it — clients reconnect and continue with a brief blip. Gotcha: automatic failover risks split-brain if the “failed” primary is actually just partitioned and still alive (now two primaries), which is why safe failover needs quorum or fencing; and failing over is only half the job — failing BACK cleanly, and not losing writes in the gap, is the harder part.