Insights:Backup and Disaster Recovery: Clarify These 6 Concepts First
Concept 1: RPO (How Much Data Can Be Lost)
RPO (Recovery Point Objective) answers the question: "How much data loss time can we tolerate?"
- Daily backup: RPO could be up to 24 hours.
- Hourly backup: RPO is about 1 hour.
- Real-time replication: RPO is close to zero, but at higher cost and complexity.
RPO is driven by business requirements, not tools.
Concept 2: RTO (How Long Recovery Takes)
RTO (Recovery Time Objective) answers the question: "How long can the business afford to be down?"
- Brochure site: A few hours may be acceptable.
- Order-processing system: Usually measured in minutes.
- Transaction or production system: Requires hot standby and failover plans.
Shorter RTO means higher investment.
Concept 3: Full vs. Incremental
- Full backup: Complete copy; simple to restore but storage-heavy.
- Incremental backup: Backs up only changes; saves space but recovery requires replaying the chain.
- Differential backup: Falls between the two.
A common combination is periodic full backups paired with frequent incremental backups.
Concept 4: Snapshots Are Not Backups
Snapshots are fine for short-term rollbacks, but they share the same failure domain as the original storage. If the source is deleted, hit by ransomware, or damaged, the snapshots may go with it.
Takeaway: Snapshots can serve as a first layer, but they cannot replace off-site backups.
Concept 5: Off-Site and Offline Copies
- In-datacenter replica: protects against single-server failures, not data-center-level incidents.
- Off-site replica: protects against regional outages.
- Offline / immutable copy: protects against ransomware and accidental deletion.
Maintain at least one copy outside the permissions of the production account.
Concept 6: Retention Policy and Recovery Drills
Retention duration depends on compliance and business needs:
- Daily backups retained for 7–30 days.
- Weekly backups retained for 1–3 months.
The crucial point is not merely "having a backup," but "having practiced recovery." Backups that haven't undergone recovery drills cannot be considered usable.
Checklist
| Question | Related Concept | Specify Clearly |
|---|---|---|
| How much data can we afford to lose? | RPO | Minutes / Hours |
| How long can the system stay down? | RTO | Minutes / Hours |
| What needs backing up? | Scope | Database, files, configurations |
| Where to store backups? | Off-site / Offline | Different fault domains |
| How long to retain? | Retention | Days / Weeks / Months |
| Can it actually be recovered? | Drill / Test | Who runs it and how often |
Boundaries
Backup and disaster-recovery solutions cannot guarantee zero business interruption, but they can contain losses within acceptable limits. Design the plan according to business value—don't apply a single strategy across all systems.
Related Reading: