Oracle Active Data Guard Architecture: Zero Data Loss, Far Sync, & Fast-Start Failover

Oracle Data Guard supports three distinct operational protection modes tailored to business RTO (Recovery Time Objective) and RPO (Recovery Point Objective) tolerances:

• Maximum Performance (ASYNC): Redo data is shipped asynchronously by the NSS background process. Zero impact on primary commit times, but allows minimal RPO lag during catastrophic sudden failures.
• Maximum Availability (SYNC / NOAFFIRM or AFFIRM): Synchronous redo transport. Guarantees zero data loss as long as the standby or network is operational; gracefully downgrades to ASYNC if the standby becomes unreachable.
• Maximum Protection (SYNC / AFFIRM): Absolute zero data loss guarantee. If redo cannot be acknowledged by at least one standby, the primary database instance halts immediately.

When primary and secondary datacenters are separated by 1,000+ miles, synchronous network round-trip times (RTT 30-50ms) degrade application commit throughput. Oracle Far Sync solves this through lightweight proxy instances located within 50 miles of the primary site.

The primary commits synchronously to the nearby Far Sync instance (0 RPO). The Far Sync instance then ships the redo asynchronously across the WAN to the remote standby datacenter without burdening the primary application.

Active Data Guard licenses permit opening the physical standby database in READ ONLY WITH APPLY mode. Heavy analytical queries, audit reports, and RMAN backups are offloaded to the standby while redo apply continues in real time without pausing replication.

During an unexpected outage, human decision latency often exceeds technical failover duration. Oracle Data Guard Broker with Fast-Start Failover (FSFO) and observer daemons automatically detects primary failure and orchestrates an orderly, zero-loss failover in under 30 seconds.