Replication across three availability zones, daily snapshots, nightly off-server backups and documented restores.
01Protection layers
| Layer | Frequency | Retention |
|---|---|---|
| Database replication across three availability zones | Continuous | Live |
| Encrypted snapshots of every server and data volume | Daily | 7 days |
| Full database backup to separate encrypted storage | Nightly | Rolling archive |
02Objectives
- Recovery point objective (RPO): at most 24 hours of data for a full restore from backup; zero for the loss of a single database node, which replication absorbs by failing over automatically.
- Recovery time objective (RTO): service restored within 24 hours of a regional-level failure, and within minutes for the loss of a single database node.
03Recovery procedure
- Restore runbooks are documented for the database, server volumes and the e-mail archive.
- Servers are rebuilt from Terraform templates; their data volumes are protected from deletion and reattach to the new host.
- Each release keeps its predecessor on the server, so a faulty deploy is rolled back by switching versions.
- After recovery the service is verified in production and the incident review records what was restored and from which point.
04Roles and communication
The incident lead coordinates recovery, engineering executes the runbooks, and customers receive updates through their account contacts until the service is confirmed healthy.
05Contact
Questions about this policy, security questionnaires and Data Processing Agreement requests: privacy@mentornity.com.