Service Uptime
Definition
Service Uptime
Service uptime is the share of an agreed period during which a service was running and reachable. It is operational minutes divided by scheduled minutes, and it is the figure most contracts attach service credits to when it falls short.
Uptime and availability are not synonyms. Uptime asks whether the service was running; availability asks whether the people relying on it could get their work done.
That distinction settles arguments — a platform can report 99.98% uptime in the same month a broken authentication service locked users out for six hours.
Exclusions do the rest of the work — planned maintenance windows, third party outages, and force majeure all carve time out of the denominator.
Read the exclusion list before the percentage. Two providers quoting 99.9% can be offering materially different promises.
Key takeaways
- Service uptime divides operational minutes by scheduled minutes in a period.
- Exclusions and measurement points matter more than the headline percentage.
- 99.9% still permits roughly 43 minutes of downtime each month.
- Uptime describes running; availability describes usable.
How it works
Agree the measurement window, agree what counts as down, then divide the minutes the service was running by the minutes it was supposed to be. Define the probe location too, because a check inside the data centre sees a healthier service than a customer does.
Nines are shorthand, not precision — each additional nine cuts permitted downtime by roughly a factor of ten, and the cost of getting there rises far faster than that.
| Uptime target | Downtime per month | Typical setting |
|---|---|---|
| 99.0% | About 7 hours 18 minutes | Internal tools |
| 99.9% | About 43 minutes | Standard managed service |
| 99.95% | About 22 minutes | Customer facing platforms |
| 99.99% | About 4 minutes 20 seconds | Payments and trading |
Standards bodies insist the definition travels with the number. NIST SP 500-307 sets out that a cloud service metric needs a definition, a unit, and stated measurement rules before it means anything.
Continuity planning sits underneath the target. NIST SP 800-34 Revision 1 gives the contingency planning method deciding whether a missed nine is a bad month or a recoverable incident.
Partial degradation is the honest gap. A service running at a tenth of its normal speed reports as up while everyone using it would call it down.
Examples
Service uptime looks straightforward until you compare how two providers measure it, and the differences usually sit in exclusions rather than engineering. Four cases show how identical percentages describe different levels of service.
A managed hosting provider. Reported 99.97% for 2024. Excluding a four hour planned window each month, the customer measured 99.62% from outside the estate.
A payments platform. Held 99.99% by running active redundancy in two regions. The second region cost roughly a third of the platform budget.
A Philippine BPO client portal. Uptime read a healthy 99.95% while agents lost two hours to a slow database. Latency was added to the definition the following quarter.
A retail ordering service. A third party payment outage cost 90 minutes. It was contractually excluded, so no credit triggered even though customers could not buy.
Related terms
Service uptime is one member of a small group of measures describing whether a system is there when needed, and each neighbour below answers a slightly different version of that same question.
- Uptime Percentage: the same measure expressed as a contract target.
- System Availability Rate: whether the service was usable rather than merely running.
- Operational Availability: the readiness view used in engineering work.
- Mean Time to Repair: how quickly a failure is put right.
- Mean Time Between Failures: how often failures arrive in the first place.
- Service Level Agreement (SLA): the contract holding the uptime commitment.
- Business Continuity Plan (BCP): the response when the target is badly missed.
FAQ
What is a good service uptime target?
99.9% suits most managed services, and 99.99% is reserved for payments, trading, and safety critical systems. Each extra nine multiplies the cost of the supporting architecture.
Is uptime the same as availability?
No. Uptime says the service was running; availability says people could use it. A login failure breaks availability while uptime stays perfect.
Does planned maintenance count as downtime?
Usually not, provided it falls inside an agreed window with notice. Check the wording, because generous windows quietly inflate the reported figure.
How much downtime does 99.9% allow?
Roughly 43 minutes a month, or about 8 hours 45 minutes across a year. That is enough for one bad incident.
Where should uptime be measured from?
From outside the estate, at the point the user sits. Internal probes consistently report a better number than customers experience.
What happens when the target is missed?
Service credits usually apply. They rarely cover the commercial damage, which is why continuity planning matters more than the credit.
Find managed service and infrastructure partners in the Outsource Accelerator BPO hubs.
Independent
