Disaster recovery (DR)

Backup = a copy of your data (and sometimes config) at a point in time. You take backups regularly (daily, hourly) so you can restore if data is lost or corrupted.

Restore = bring back data or a system from a backup. RTO says how fast you need to be back; RPO says how much data loss you accept. Backups and restore procedures must meet those targets.

Cloud: automated backups (RDS, EBS snapshots), versioning (S3), and cross-region replication. Test restore regularly-backups you never test may fail when you need them.

Define RTO and RPO. Design backups and replication for RPO; design failover and restore for RTO. Run DR drills.

Multi-region = run (or replicate) in more than one geographic region. If one region goes down, you fail over to another. Used for DR and sometimes for low latency (serve from the region nearest the user).

Replication = copy data to another region or AZ. Sync replication = strong consistency, higher latency. Async replication = eventual consistency, lower latency. Choose based on RPO and application needs.

Define RTO and RPO. Design backups and replication to meet RPO. Design failover and restore procedures to meet RTO. Run DR drills (restore from backup, failover to another region) so the process works when it matters.