Know RPO (Recovery Point Objective)

Severity: Recommended

RPO is the maximum acceptable data loss measured in time. It drives backup frequency and replication strategy.

Check automatically:

1. Look for RPO documentation:

# Search for RPO mentions
grep -riE "RPO|recovery.*point.*objective|data.*loss|backup.*frequency|point.*in.*time" docs/ runbooks/ README.md CLAUDE.md SLA* --include="*.md" 2>/dev/null

2. Check backup frequency:

# Check for backup schedules
grep -riE "backup.*schedule|cron.*backup|daily.*backup|hourly.*backup" .github/ scripts/ terraform/ --include="*.yml" --include="*.tf" --include="*.sh" 2>/dev/null

# Check for point-in-time recovery (PITR)
grep -riE "point_in_time|pitr|continuous.*backup" terraform/ infrastructure/ --include="*.tf" 2>/dev/null

RPO tiers and required strategies:

RPO	Strategy Required
0 (no data loss)	Synchronous replication, multi-region writes
< 1 min	Async replication, streaming WAL
< 1 hour	Point-in-time recovery (PITR)
< 24 hours	Daily backups
< 1 week	Weekly backups

Ask user:

• "How much data loss is acceptable? (1 hour? 1 day?)"
• "What's your backup frequency?"
• "Do you have point-in-time recovery enabled for your database?"
• "Is RPO agreed with stakeholders/business?"

Cross-reference with:

• RR-007 (RTO) - often defined together as recovery objectives
• RR-005/RR-006 (recovery docs/testing) - RPO should be mentioned and validated
• Section 26 (backups) - backup frequency determines achievable RPO

Pass criteria:

• RPO is defined (even informally: "losing a day of data would be bad")
• Backup frequency supports the RPO (daily backups = 24h RPO max)
• Team understands the tradeoff (tighter RPO = higher cost)
• For critical data: PITR enabled or frequent backups

Fail criteria:

• No idea what acceptable data loss is
• Backup frequency doesn't match expectations (weekly backups but expect no data loss)
• RPO defined but infrastructure doesn't support it
• Never verified backup restore point (might be older than expected)

Evidence to capture:

• Defined RPO (or lack thereof)
• Actual backup frequency
• Whether PITR is enabled
• Gap between target RPO and actual capability

---