A reference architecture for running fault-tolerant, high-availability enterprise workloads across multiple OCI regions, including traffic management, data replication, and failover automation.
Topology overview
Two active OCI regions in an active-active configuration for stateless application tiers, with a third region reserved for cold-standby database recovery — chosen to balance resilience against the operational cost of running a full third active region.
Traffic Management Steering Policies route users to the nearest healthy region based on geoproximity, with automatic failover on health-check failure within 30 seconds.
Data layer design
Autonomous Database with cross-region Data Guard provides synchronous replication within a region and asynchronous replication to the DR region, keeping documented RPO under 5 seconds for the primary pair and under 5 minutes for the DR region.
Object Storage cross-region replication handles static assets and backups independently of the database replication path, avoiding a single point of coordination failure.
Operational considerations
Runbooks for both planned failover (maintenance windows) and unplanned failover (regional outage) are maintained separately, since planned failover can afford a controlled drain of in-flight requests that unplanned failover cannot.
Cost model: running the second active region roughly doubles compute spend for the application tier, but the database DR region costs significantly less since it runs standby-sized shapes until an actual failover event.
Leave a Reply