Mean Time To Repair (MTTR): Measuring Your Fix Velocity
MTTR measures how quickly your team can fix a problem once active work begins. It's the 'wrench time' of incident response, not total outage duration. SREs track it to gauge runbook and diagnostic effectiveness.
WHY IT EXISTS Systems inevitably fail. To improve reliability, we need to measure not just that they failed, but how efficiently we can fix them. MTTR was created to isolate and measure the performance of the repair process itself, separate from detection, queuing, or supply chain delays.
THE MENTAL MODEL Think of a pit crew in a race. The total time the car is off the track is the total outage. MTTR is only the time the crew is actively working on the car—changing tires, fixing a wing. It does not include the time it took the car to limp back to the pit lane or time spent waiting for a special tool to be brought from the main garage. It measures the crew's direct performance.
HOW IT WORKS MTTR is calculated with a simple formula: Total Corrective Maintenance Time divided by the Total Number of Failures over a given period. For instance, if you had 3 failures in a month with active repair times of 30, 60, and 20 minutes, the total repair time is 110 minutes. The MTTR would be 110 minutes / 3 failures, which is approximately 36.7 minutes.
WHEN TO USE IT Use MTTR to measure the maintainability of a system and the effectiveness of your response team. It is a core Key Performance Indicator (KPI) for Site Reliability Engineering (SRE) and operations teams. A decreasing MTTR trend signals that your team's diagnostic skills, runbooks, and tooling are improving.
WHEN NOT TO USE IT Do not use MTTR as the sole measure of customer impact. It only captures the active repair phase. It explicitly excludes time spent waiting for parts, administrative approvals, or other logistical delays (known as Administrative or Logistic Downtime). Confusing MTTR with total outage time hides inefficiencies in other parts of your incident lifecycle.
ONE CANONICAL EXAMPLE A database server fails. An alert fires at 10:00 AM. An engineer begins actively diagnosing the issue at 10:05 AM. They identify a corrupted configuration file and deploy a fix. The database is back online at 10:25 AM. The time to repair for this incident is the period of active work: from 10:05 AM to 10:25 AM, which is 20 minutes. This 20-minute duration is one data point for calculating the overall MTTR.
Read the original → en.wikipedia.org
Get five bites like this every day.
Tezvyn delivers a daily feed of 60-second tech bites with quizzes to lock in what you learn.