Planning server backups: The 3-2-1 rule, retention and restore tests

Snapshots and backups solve different problems

A snapshot is useful for preserving a known state before an update. If it lives on the same storage as the server, however, it cannot protect against that storage failing. A backup is an independent copy with its own retention. Important services often need both: quick rollback and recovery after a larger incident.

Use the 3-2-1 rule as a starting point

Keep three copies of the data on two different storage types, with one copy physically or logically separated. It is a framework rather than a rigid law. The essential point is that one account, operator mistake or compromised machine must not be able to erase every copy.

RPO and RTO in plain language

• RPO: How much data can be lost between the latest backup and an incident?
• RTO: How long can recovery take before the service must be available?

A private test server may be fine with a daily backup. A shop or community database may need more frequent copies and a defined restart plan. These targets determine both the technology and the budget.

Different data needs different treatment

Databases should be exported consistently or protected with a database-aware method. Include configuration, uploads, certificate information and an inventory of installed services. Encrypt stored secrets. Regularly verify that new volumes and applications have not been left outside the backup scope.

Put recovery tests on the calendar

Restore a backup into an isolated environment on a fixed schedule. Measure the duration, inspect logs and test a real user action. Document manual steps until the emergency process becomes repeatable. Alert when a backup is missing or unexpectedly small.

Read more about Proxmox-based protection under Proxmox Backup Server, or explore suitable compute under server packages.