Database Backup & Disaster Recovery (RPO vs RTO)

Every disaster recovery (DR) architecture is defined and evaluated by two critical metrics:

1. Recovery Point Objective (RPO) — Data Loss Window

RPO measures the maximum acceptable amount of data loss expressed in units of time. It answers: "How much committed data can the business afford to lose in a catastrophe?"

• RPO = 24 hours: Database is backed up once per night. If the primary storage array catches fire at 11:59 PM, 23 hours and 59 minutes of customer transactions are permanently lost.
• RPO = 1 minute: Database streams Write-Ahead Logs (WAL) to remote object storage every 60 seconds. At most 60 seconds of recent writes are lost during a sudden data center outage.
• RPO = 0 (Zero Data Loss): Synchronous multi-region replication ensures transactions are committed to at least two distant regions before returning success to the client.

2. Recovery Time Objective (RTO) — Downtime Window

RTO measures the maximum acceptable elapsed duration between an outage and the full restoration of service. It answers: "How long can the application stay offline?"

• RTO = 4 hours: A replacement database instance must be provisioned, the multi-terabyte backup downloaded from S3, and WAL logs replayed before traffic can resume.
• RTO = 30 seconds: An automated orchestrator (e.g., Patroni / AWS Aurora failover) detects primary node failure and promotes an existing warm standby replica to primary almost instantly.