What is SQL Server Always On?
SQL Server Always On is the native high availability and disaster recovery feature set introduced in SQL Server 2012. It has two parts: Always On Availability Groups (AGs), which replicate databases to readable secondary replicas without shared storage, and Always On Failover Cluster Instances (FCIs), which fail over an entire SQL instance across shared storage. For most modern estates, AGs are the preferred path because they need no shared storage, support readable secondaries and protect groups of databases together.
Always On Availability Groups sit on top of a Windows Server Failover Cluster (WSFC). Get the cluster right first: node quorum, a file-share or disk witness, and dedicated heartbeat networks. A flaky cluster foundation will take down even a perfectly configured AG.
SQL Server high availability options: AGs, FCIs and mirroring
Choosing a high availability strategy means matching the technology to the workload, edition and recovery objectives. Database mirroring is deprecated and limited to a single secondary. Failover Cluster Instances protect a whole instance but need shared storage, so they do not protect against a storage failure. Availability Groups replicate at the database level, need no shared storage, and support up to eight secondaries — synchronous and asynchronous — with readable replicas and multi-database failover.
Edition matters: full Always On Availability Groups are available on SQL Server Enterprise edition, while SQL Server Standard edition supports basic availability groups (a single database per AG) and FCIs. For any SQL Server 2012+ Enterprise estate that needs RPO near zero on the primary data centre and a warm DR replica off-site, AGs are the right tool.
Synchronous vs asynchronous replicas
Synchronous-commit replicas give zero data loss (RPO=0) but add commit latency — keep them within a low-latency, same-data-centre link. Asynchronous-commit replicas accept a small RPO in exchange for cross-site resilience; use them for the DR replica in a secondary region.
- Primary + 1 synchronous secondary in the primary DC
- 1 asynchronous secondary in the DR DC for disaster recovery
- Readable secondaries for reporting / backup offload
- Quorum witness (file share or disk) to prevent split-brain
Backup strategy on AGs
Configure backup preferences — prefer secondaries — so backups run on a readable replica and offload the primary. Transaction log backups are still the backbone of point-in-time recovery; schedule them every 5–15 minutes for busy OLTP databases. Test restore to the DR replica quarterly so the runbook is proven, not theoretical.
Lessons learned
- Build a solid WSFC foundation before configuring the AG
- Use synchronous commit in-DC, asynchronous commit to DR
- Offload backups and reporting to readable secondaries
- Run a forced-failover drill quarterly and time the recovery
Written by
Jamshed Alam
Global MSP & IT Infrastructure Consultant with 20 years across Exchange, Microsoft 365, Azure, security and automation. Available for project and contract work worldwide.