An uptime calculator turns an availability percentage, such as 99.9 percent, into the amount of downtime that it allows each day, week, month and year. It also works the other way: enter the downtime that you had to find your availability, or enter the mean time between failures and the mean time to repair. A fourth mode combines the availability of parts that are connected in series or in parallel. It is useful for service level agreements, hosting and system design. It runs in your browser.
How to use the uptime calculator
- Choose downtime allowed, availability from downtime, MTBF and MTTR or series and parallel.
- Enter the numbers.
- Press Calculate.
What availability is
The Wikipedia article on availability defines it as uptime divided by the total time, which is uptime plus downtime.[2] For a repairable item, the inherent availability is MTBF ÷ (MTBF + MTTR), where MTBF is the mean time between failures and MTTR the mean time to repair.[2] A system that fails once every 999 hours and takes 1 hour to repair has an availability of 999 ÷ 1,000, which is 99.9 percent. The calculator uses these formulas, and shows the downtime per year that follows.
The nines
Availability is often described by the number of nines. The article on high availability has a table of the downtime for each level. At 90 percent, the downtime is about 37 days a year. At 99 percent it is 3.7 days a year, about 7.3 hours a month and 14 minutes a day. At 99.9 percent it is 9 hours a year, 44 minutes a month, 10 minutes a week and 1.4 minutes a day. At 99.99 percent it is 53 minutes a year, 4.4 minutes a month, 1 minute a week and 8.6 seconds a day. At 99.999 percent it is 5.3 minutes a year, 26 seconds a month, 6 seconds a week and 860 milliseconds a day.[1] The calculator gives these figures with more digits, and for any availability that you enter.
How the downtime is worked out
The downtime is the unavailability, 1 minus the availability, times the length of the period. The page states a rule of powers of ten: the downtime per day is 8.64 × 10^(4 − n) seconds for n nines.[1] For three nines, that is 86.4 seconds a day. The pages do not say how long a year or a month is, but the table agrees with a year of 365 days, which has 8,760 hours, and a month of one twelfth of that, so the calculator uses them. A table that uses 365.25 days gives 8.77 hours for three nines, instead of 8.76.
Series and parallel systems
When parts are connected so that all of them must work, the availability is the product of the availabilities of the parts, and it is always lower than that of any one of them. When parts are redundant, so that one is enough, the availability is 1 minus the product of the unavailabilities. For N identical parts, that is 1 − (1 − X)^N.[2] The page's example is ten hosts that are each only 50 percent available, which together reach 99.9023 percent when any one is enough. Enter ten values of 50 in the parallel mode to see this. The page stresses that this assumes that the parts fail independently, for example by not being in the same data centre.[2]
Two 99.9 percent parts
Two services of 99.9 percent that must both work have a combined availability of 99.8001 percent, which is about 17.5 hours of downtime a year, nearly twice as much as one part. Two such services in parallel give 99.9999 percent, which is about half a minute a year. The calculator shows both, and the number of nines. This is why a chain of many dependent services has a lower availability than any of them.
Reading a service level agreement
A promise of 99.9 percent uptime means up to about 44 minutes of downtime in a month. Check what the agreement counts: whether planned maintenance is excluded, how the month is measured, and what the credit is if the provider misses the target. The calculator gives the raw downtime, and the terms of the agreement may reduce it. Compare the figure with what your business can tolerate, since each extra nine is much harder and costlier to achieve.
Planning for availability
Work out how much downtime the business can accept, and choose the availability to match. A site that can be down for a few hours a month needs about 99.5 percent, while a payment system may need 99.99 percent. Then look at the weak parts. The availability of a chain is lower than that of its weakest link, so improving one very good part gives little, and adding a second copy of the weakest part, in parallel, often gives the most. The calculator lets you test each change quickly.
Common mistakes
- Mixing up MTBF and MTTF. MTBF is for repairable systems.
- Counting only unplanned downtime. Planned outages may count in some agreements.
- Assuming independence. Parts in the same data centre often fail together.
- Adding availabilities. Multiply them for series parts.
What next
Use this often? Add Uptime to your Home screen.
Frequently asked questions
What is an uptime calculator?
A tool that turns an availability percentage into the downtime it allows per day, week, month and year, and the other way round.
How much downtime does 99.9 percent allow?
About 8.76 hours a year, 43.8 minutes a month and 1.44 minutes a day.
How much downtime does 99.99 percent allow?
About 52.6 minutes a year, 4.38 minutes a month and 8.64 seconds a day.
What is the availability formula?
Uptime ÷ (uptime + downtime), or MTBF ÷ (MTBF + MTTR) for a repairable item.
How do I combine the availability of several parts?
In series, multiply the availabilities. In parallel, use 1 minus the product of the unavailabilities.
Was this tool helpful?
What went wrong?
We only receive the reason you pick, never what you typed into the tool.
Thanks for your feedback.