4.7.2. Restore Methods and Backup Validation
💡 First Principle: The restore process must be understood before the backup is designed, not after a failure. Restore decisions—where to restore, whether to overwrite, how to handle conflicts—must be planned for in advance because making them under pressure during an actual failure leads to errors.
Restore Methods
| Method | What It Does | When to Use |
|---|---|---|
| Overwrite | Restores data over existing files (replaces current with backup) | Clean recovery after failure; current data is bad |
| Side by side | Restores to same location but renames existing files | Comparison; selective recovery; verify before replacing |
| Alternate location path | Restores to a different folder or server | Testing; don't want to affect production; partial recovery |
Backup Validation
A backup job that "completes successfully" is not validated until a restore has been tested. Validation elements:
- Media integrity: Can the backup media be read? (Tape verification, backup catalog integrity checks)
- Equipment: Is the restore hardware compatible? (Tape drive generations, encryption key availability)
- Regular testing intervals: Schedule restore tests—monthly for critical systems, quarterly for others. Restore a random file, restore a full system to a test environment.
- Media inventory before restoration: Before starting a restore, verify all required media (all incrementals in a set, for example) is present and readable.
The Critical Validation Point
If you've never successfully restored from your backups, you don't know they work. Common failure modes discovered only during restore testing:
- Backup encryption keys not documented (can't decrypt backup)
- Tape drive read failures (different hardware than backup hardware)
- Application-consistent backup not configured (database backup is inconsistent, restores but won't start)
- Missing incrementals (one day's backup was skipped unnoticed)
⚠️ Exam Trap: "Backup validation" means testing restore, not just verifying the backup job completed without errors. A job that writes to tape successfully but produces an unreadable tape appears successful in logs. Only a restore test catches this failure.
Reflection Question: An administrator sets up automated nightly backups and receives a "Backup Completed Successfully" email every morning for 6 months. Is the backup strategy validated? What specific action would actually validate it?
