AWS backup strategy layers automated snapshots, cross-region copies, and application-consistent exports. Test restores—untested backups are wishful thinking.
Service-specific backups
- RDS/Aurora — automated backups, manual snapshots, PITR
- EBS — snapshots to S3 (managed)
- S3 — versioning, cross-region replication, Glacier lifecycle
- DynamoDB — on-demand and PITR backups
- AWS Backup — centralized policies across services
RTO and RPO
RPO (Recovery Point Objective)—max acceptable data loss. RTO (Recovery Time Objective)—max downtime. Backup frequency and Multi-AZ design derive from these targets.
Create RDS snapshot
aws rds create-db-snapshot \
--db-instance-identifier my-learning-db \
--db-snapshot-identifier my-learning-db-$(date +%Y%m%d)Practice: Use RDS and DynamoDB free tier where eligible. Delete sandbox databases when finished—storage and instance hours can incur charges.
Run a restore drill in sandbox quarterly—even for learning accounts—to understand steps and timing.
Important interview questions and answers
- Q: PITR?
A: Point-in-time recovery restores RDS to any second within retention window using transaction logs. - Q: Why restore drill?
A: Proves backups work and documents actual RTO—not theoretical hope.
Self-check
- What do RPO and RTO mean?
- Name two AWS backup mechanisms for RDS.
Pitfall: Snapshots you never restored do not prove you can recover—schedule a restore drill.
Interview prep
- RPO vs RTO?
RPO is max data loss window; RTO is max acceptable downtime.
- Restore drill?
Proves backups work and defines real recovery time objective.