Mean Time to Repair
Definition
Mean Time to Repair
Mean time to repair is the average time taken to restore a failed system to working order, measured from when repair work starts to when service returns. It is the recovery half of reliability, and it moves availability faster than failure frequency does.
The clock boundaries decide the number. Repair time proper excludes detection and waiting — stretch it to cover both and you are measuring something else entirely.
Availability depends on it directly. Halving repair time improves uptime as much as halving the failure rate, and it is usually far cheaper to achieve.
Key takeaways
- Mean time to repair divides total repair time by the number of repairs completed.
- The clock covers active restoration work, not detection or waiting for parts.
- Repair time and failure frequency together determine availability.
- Spares, documentation, and access rights cut repair time more than skill does.
How it works
Mean time to repair is calculated by dividing the total time spent actively restoring failed systems during a period by the number of repairs completed in that same period, giving an average recovery duration.
The formula is: total repair time ÷ number of repairs.
The intervals around repair are separate measures, and mixing them is the most common reporting error.
| Interval | Clock runs from | Clock stops at |
|---|---|---|
| Detect | Incident start | Awareness of the incident |
| Respond | Awareness | First corrective action |
| Repair | Work begins | Service restored |
| Resolve | Incident start | Root cause removed |
Row three is this metric. Rows one, two, and four belong to detection, response, and resolution measures respectively, and reporting a blended figure hides which stage is actually slow.
Reliability theory frames the pairing precisely. The NIST/SEMATECH e-Handbook explains why product reliability matters commercially, noting that repeated failure produces lasting customer dissatisfaction that damages market position.
The underlying property has a formal definition. The American Society for Quality defines reliability as the probability that a product, system, or service performs its intended function adequately.
The same definition also covers operating without failure in a defined environment.
Preparation beats expertise. Available spares, current runbooks, and pre-granted access rights shorten repair — far more reliably than hiring senior engineers does.
Waiting time deserves its own line. Parts delivery and vendor response are real delays, and burying them inside repair time hides a supply problem as an engineering one.
The measure belongs inside the service level agreement (SLA), where restoration targets are usually tiered by severity.
Never average across severities. A five-minute password reset and a two-day database rebuild are not the same event, and averaging them describes neither.
Examples
Repair performance depends on preparation, access, and how much redundancy stands between a failure and the customer. Five cases show what actually shortens the clock.
Cloud platforms repair by replacing rather than fixing. Failed instances are terminated and rebuilt automatically, which pushes repair time towards seconds.
Manufacturers repair against spares inventory. Holding critical parts on site converts a three-day wait into a two-hour job, which is a stocking decision rather than a maintenance one.
Hospitals tier repair targets by clinical risk. Imaging equipment carries hours while administrative systems carry days, and the contract says so explicitly.
Retail chains repair point-of-sale hardware by swap. Field engineers carry replacement units, so the store is trading again long before the fault is diagnosed.
Outsourced infrastructure teams report repair time by severity — buyers should confirm whether vendor waiting time sits inside or outside the clock, since that single choice can halve the reported figure.
Related terms
Mean time to repair sits in the middle of the incident timeline, between detection and full resolution. The terms below cover the roles, the tooling, and the contracts that surround it.
- Site Reliability Engineer: the role that designs systems to recover quickly.
- Technical Support Engineer: the role performing most restoration work.
- Application Support Engineer: the role restoring software-layer failures.
- IT Support Technician: the front line that logs and escalates failures.
- Ticketing System: the tool supplying the timestamps this metric depends on.
- Help Desk Support: the function that owns communication during a repair.
- Service Level Agreement (SLA): the contract that tiers repair targets by severity.
FAQ
How is mean time to repair calculated?
Divide the total time spent actively restoring failed systems by the number of repairs completed in the same period.
Does the clock include waiting for parts?
It should not. Track vendor and parts delays separately, or a supply problem will be reported as an engineering one.
How does repair time affect availability?
Directly. Halving repair time improves uptime as much as halving the failure rate, and usually costs less.
What shortens repair time most?
Spares availability, current runbooks, and pre-granted access rights, ahead of additional technical skill.
Should repairs of all severities be averaged?
No. Tier the measure by severity, since a password reset and a database rebuild are not comparable events.
How does it differ from mean time to resolve?
Repair stops when service returns, while resolve continues until the root cause is removed.
Source partners contracting on restoration targets can compare delivery models via Outsource Accelerator hubs.
Independent
