In 30 seconds
- Measures how fast a team recovers from incidents.
- A DORA stability metric.
- Fast detection and easy rollback drive it down.
How it's measured
MTTR = total downtime / number of incidents, measured from when an incident starts (or is detected) to when service is restored. DORA calls it time to restore service.
Why recovery speed matters
Failures will happen. Teams that recover in minutes can deploy more boldly than teams that take days. It balances change failure rate in the DORA metrics.
Recovering faster
- Good monitoring and alerts to detect problems quickly
- One-click rollback or feature flag kill switches
- Clear incident roles and runbooks
- Blameless post-incident reviews
Myth vs reality
Myth · MTTR is about fixing the root cause.
Reality · It's about restoring service; root cause can come later.
Myth · A low MTTR means few incidents.
Reality · It measures recovery, not frequency.
Myth · Averages tell the full story.
Reality · One long outage can hide behind many quick ones.
FAQ
What does the R stand for?
Usually recovery or restore; sometimes repair or resolve. Define it clearly.
What's a good MTTR?
DORA's top performers have reported restoring service in under an hour.
Should we use median instead?
Median plus the worst cases often gives a clearer view.