MTTR (Mean Time to Recovery), also called Mean Time to Restore, measures the average duration required to recover from a system failure or service outage. It spans the entire incident lifecycle: from the moment a failure occurs through detection, notification, diagnosis, remediation, and verification that service is fully operational again.
MTTR is calculated by dividing the total recovery time across all incidents by the number of incidents. For example, if four incidents in a month took 8, 23, 45, and 12 minutes to resolve, the MTTR would be (8 + 23 + 45 + 12) ÷ 4 = 22 minutes.
The metric encompasses several sequential phases:
| Phase | Meaning |
|---|---|
| Detection | Time between failure and monitoring system alert |
| Notification | Time for alert to reach the on-call responder |
| Diagnosis | Time to identify the root cause |
| Remediation | Time to apply a fix (rollback, restart, config change) |
| Verification | Time to confirm service has returned to normal |
MTTR is one of the core DORA metrics used by modern DevOps teams to measure reliability and response capability. It works alongside uptime monitoring and health checks to track system stability.
WarningDetection often accounts for the largest portion of MTTR. A monitoring system checking every 5 minutes adds an average 2.5-minute delay compared to one checking every 30 seconds. MTTR measures incident duration, not frequency—a team with fast recovery but frequent incidents still delivers an unreliable experience.