Glossary · Reliability metrics
Mean time to repair
MTTR is the average time it takes to restore a service after a failure, from the moment it breaks to the moment it works again. The R can stand for repair, recovery, restore or resolve, so check which one a report uses.
MTTR covers detection, response, diagnosis, the fix and checking that it worked. Bringing it down is often about finding out sooner and reaching the right person faster, not only fixing faster: a 5-minute fix after 40 minutes of nobody noticing is a 45-minute incident.
Mean time to recovery ends when users are served again; mean time to resolve can run until the underlying cause is fixed, which may be days later. Say which one you report.
Averages hide spread: one long outage can double a month’s MTTR. Look at the individual incidents too, and at MTTA, mean time to acknowledge, to see how much of the time passed before anyone started working. MTTD is the detection part.
MTTR = total time to restore ÷ number of incidents
Example: Three incidents lasting 12, 30 and 18 minutes: (12 + 30 + 18) ÷ 3 = 20 minutes.
The incidents page in Uptimecheck shows time to resolve (MTTR) and time to acknowledge (MTTA) over the last 30 days. Each incident runs from the first failed check to recovery, with a timeline of checks, alerts, acknowledgements and notes, and the evidence from when it opened: the failing response, and a screenshot when the server answered or a traceroute when it didn’t.
14-day free trial · No credit card required
Guides 6 min read
MTTR, MTTA and MTBF measure how fast you fix incidents, how fast you respond and how often things break. Formulas, a worked example and practical ways to improve each one.
ReadGuides 6 min read
A copy-ready incident postmortem template, the blameless principles behind it, a filled-in example from a checkout outage and how to publish a short summary on your status page.
ReadIt depends on the service and on how incidents are counted, so comparing with other companies says little. Track the trend for your own service: most outages detected within minutes and resolved within an hour is a reasonable aim for a web service.
MTTA, mean time to acknowledge, ends when someone takes the incident. MTTR ends when the service works again. MTTA is the response part of MTTR.
Detect sooner with frequent checks, send alerts that reach the person on call, include evidence in the alert so diagnosis starts at once, and keep runbooks for failures you have seen before.
Set up your first monitor in under a minute, or look around the live demo first.
14-day free trial · No credit card required
With your OK, Google Analytics and Microsoft Clarity show us which pages help and where visitors get stuck. They set cookies for up to 13 months. Without your OK, neither of them runs. You can change your mind at any time under Cookie settings at the bottom of the page. Privacy policy