Mean time to restore
Mean time to restore (MTTR, also called 'Time to Restore Service' in the 2024 DORA report) is the average time between a production incident's detection and its resolution. It is the third DORA metric. DORA 2024 thresholds: elite = under one hour; high = between one day and one week; medium = between one week and one month; low = between one and six months.
DORA uses MTTR-as-Restore (wall-clock from incident-fired to service-restored), not MTTR-as-Repair (engineering time spent fixing the underlying bug) or MTTR-as-Respond (time to first human acknowledgement). The framing matters because Restore can be achieved by feature-flagging a bad change off, by rolling back, by failing over to a healthy region, or by hotfixing — and which of these is fastest depends on the org's CI/CD investment, on-call runbook quality, and feature-flag infrastructure. Elite-quartile MTTR is dominated by the rollback path; teams that ship behind feature flags routinely restore service in minutes regardless of how long the underlying fix takes. Distinct from the long-tail MTTR-as-Repair, which describes engineering velocity on debugging.
Related terms
- Four key metrics
The four key metrics are the DORA framework for measuring software-delivery performance: deployment frequency, lead time for changes, mean time to restore, and change failure rate.
- MTTR
Mean Time To Recovery is the average elapsed time between an incident's detection and its resolution.
- Deployment frequency
Deployment frequency is the rate at which a team deploys to production.