Lesson 6: Backup, Restore, Availability, and Disaster Recovery

Students often use "backup" and "recovery" as if they mean the same thing. They are related, but not identical.

• A backup is a copy used for restoration.
• Recovery is the process of bringing the database back to a required state after failure.

A good DBA asks not only, "Do we have backups?" but also:

• Can we actually restore them?
• How long will restore take?
• How much data can we afford to lose?
• Are logs available for point-in-time recovery?

Important backup ideas include:

• full backup — complete copy of the database
• incremental backup — captures changes since a baseline
• differential backup — captures changes since the last full backup
• logical backup — exports schema/data as logical statements or dumps
• physical backup — copies physical database files or blocks, depending on DBMS support

A backup that has never been tested is only a promise, not proof.

When organizations recover from failure, two measures are very important:

Example:

• If an enrollment system has RPO = 15 minutes, the institution can tolerate at most 15 minutes of lost data.
• If it has RTO = 2 hours, the system must be restored within 2 hours.

These values shape the backup plan. A weekly backup alone is not enough for a system with strict RPO and RTO.

Restore planning should include:

• where backup files are stored
• who is authorized to restore
• what order steps must happen
• how to verify a successful restore
• how to communicate during incident response

DBAs are expected to think in service terms, not just file terms.