RDS snapshots and retention: DR that survives audit
Retention policies, PITR, and failover tests for critical AWS databases.
RDS databases in AWS power ERP, EHR SaaS backends, and custom apps for Puerto Rico SMBs — but default 7-day snapshot retention and disabled PITR fail HIPAA audit and post-hurricane DR expectations. The AWS Security Blog recommends automated backups with retention aligned to business RPO; NIST 800-34 requires periodic recovery tests.
Configure minimum 35-day backup retention for critical prod; consider pre-deploy manual snapshots tagged Release. PITR enabled — recovery window hours based on transactionality. Cross-region snapshot copy for continental DR if PR operation depends on Miami HQ access during island emergency.
Failover testing: documented annual RDS Multi-AZ failover; for Single-AZ accepted for cost, restore snapshot in staging and validate app connectivity. Document measured RTO, not slide theoretical. Auditors ask when last successful test occurred.
Encryption: KMS CMK for snapshots, annual key rotation, restrict kms:Decrypt to backup/restore roles only. Accidentally public snapshots are GuardDuty Critical finding — IAM policy and SCP prevent sharing.
Accumulated snapshot cost: lifecycle policy deletes stale manual snapshots; AWS Backup centralizes cross-resource policy. FinOps and DR are not opposites — retention informed by legal hold and regulatory, not bill fear. Guaynabo clinic with 7-year PHI retention needs different architecture than tourism ecommerce — design per dataset.
Operations and execution — connecting strategy with what the team ships every week.
