The two-region pilot-light DR pattern operates with read/write in the primary region while continuously replicating the database and S3 objects to the secondary region, with DNS failover via Route 53 and a scripted failover that promotes the read replica and scales the dormant ASG in parallel.

Concrete RPO targets under one minute and RTO under about 30 minutes—RTO is largely driven by RDS promotion—and the pilot light approach is shown as the most cost-effective DR pattern compared with backups, warm standby, active-active, or full active/standby models.

The plan includes concrete commands and endpoints to verify failover success, such as health checks that return a region header to ensure clients are directed to the healthy region.

Overall, the story provides motivation, architecture, implementation steps, operational runbook, trade-offs, and verification steps for a two-region pilot-light DR pattern on AWS.

In this setup, eu-west-1 Ireland handles active user traffic while eu-west-3 Paris is continuously updated but idle, with Route 53, a failover script, and a single writable database promotion to recover in the secondary region, achieving measurable RTO and RPO.

The failover chain details automated health checks and promotions: local and ALB health checks, Route 53 health probes, DNS failover, CloudWatch/SNS alerts, execution of the failover script, and then promotion plus ASG scaling to restore service in the second region.

Terraform modules deploy identical regional stacks—two-AZ VPCs, TLS-terminated ALB, Auto Scaling group, RDS PostgreSQL, S3 for attachments, and Secrets Manager for DB credentials—with the primary writable database active and the other region kept dormant.

Two replication streams and a single irreversible step govern the transition: cross-region RDS PostgreSQL replica promotion to writable in the secondary region, with S3 and Secrets Manager replication maintaining availability.

The article presents a practical, multi-region disaster recovery pattern on AWS using the pilot-light approach, complete with architecture details, implementation notes, and a live-runbook.