A configuration of two or more SQL Server nodes that work together to minimize downtime by automatically detecting failures and failing over database workloads to healthy nodes.
- SQL Server high availability (HA) architectures include: Windows Server Failover Clustering (WSFC), Always On Availability Groups, and Always On Failover Cluster Instances (FCI).
- Failover Cluster Instances (FCI) treat multiple nodes as a single SQL Server instance with shared storage: They are transparent to applications through a virtual network name.
- Always On Availability Groups complement or replace FCIs: They provide per-database failover, readable secondaries, and support both local and geographically distributed replicas.
- DBAs must understand the difference between planned and unplanned failover: Planned failover supports zero data loss during scheduled maintenance, while unplanned failover is automatic and may involve data loss in asynchronous configurations.
- HA cluster health monitoring includes: Quorum state, cluster node status, replica synchronization, and network heartbeat verification.
- Recovery Time Objective (RTO) is the primary design consideration for HA cluster configurations: It defines how quickly services must be restored after a failover event.
- Relevant Idera tools: SQL Diagnostic Manager monitors cluster node health, Availability Group replica status, and failover events with configurable alerting.
- Related terms: Always On Availability Groups, Failover Cluster Instance, Quorum, RTO, RPO.
