Uptime, downtime and availability calculator
Uptime % = (total time - downtime) ÷ total time × 100. 99.9% uptime allows 43m 12s of downtime a month, or 8h 45m 36s a year; 99.5% allows 3h 36m a month. Pick a target to see the downtime it permits, enter an outage you already had to calculate the uptime it works out to, or paste a server uptime reading to convert it into days, hours and minutes. The arithmetic runs in your browser, and nothing is sent anywhere.
Common SLA targets and the downtime they allow
The uptime levels that turn up in contracts, and what each one allows. A month here is 30 days, a quarter is 91.25 days (365 ÷ 4) and a year is 365 days.
| Uptime | Also called | Per day | Per week | Per month | Per quarter | Per year |
|---|---|---|---|---|---|---|
| 99% | Two nines | 14m 24s | 1h 40m 48s | 7h 12m | 21h 54m | 3d 15h 36m |
| 99.5% | — | 7m 12s | 50m 24s | 3h 36m | 10h 57m | 1d 19h 48m |
| 99.9% | Three nines | 1m 26s | 10m 5s | 43m 12s | 2h 11m 24s | 8h 45m 36s |
| 99.95% | — | 43s | 5m 2s | 21m 36s | 1h 5m 42s | 4h 22m 48s |
| 99.99% | Four nines | 9s | 1m | 4m 19s | 13m 8s | 52m 34s |
| 99.999% | Five nines | 1s | 6s | 26s | 1m 19s | 5m 15s |
Work out your allowed downtime
A percentage between 0 and 99.9999. Try 99.9, 99.95 or 99.99.
| Period | Length | Allowed downtime |
|---|---|---|
| Per day | 24 hours | 1m 26s |
| Per week | 7 days | 10m 5s |
| Per month | 30 days | 43m 12s |
| Per quarter | 91.25 days (365 ÷ 4) | 2h 11m 24s |
| Per year | 365 days | 8h 45m 36s |
Work backwards from an outage
How long the service was actually down. The uptime it works out to depends entirely on the period you measure it over.
| Measured over | Length | Uptime |
|---|---|---|
| A day | 24 hours | 97.0139% |
| A week | 7 days | 99.5734% |
| A month | 30 days | 99.9005% |
| A quarter | 91.25 days (365 ÷ 4) | 99.9673% |
| A year | 365 days | 99.9918% |
Convert a server uptime reading
Paste what the machine printed: seconds (49000000), days:hours:minutes:seconds (114:16:45:17, as Windows Task Manager shows it), PowerShell's Get-Uptime (3.04:05:06), or the Linux uptime line (up 114 days, 16:45). A bare 15:01 is read as 15 hours 1 minute; write 0:15:01 for 15 minutes 1 second. Windows and SNMP counters are usually milliseconds or hundredths of a second, so divide those down to seconds first.
| Unit | Broken down | Whole reading in this unit |
|---|---|---|
| Days | 114 | 114.6981 |
| Hours | 16 | 2,752.75 |
| Minutes | 45 | 165,165.28 |
| Seconds | 17 | 9,909,917 |
A month is counted as 30 days, a quarter as 91.25 days (365 ÷ 4) and a year as 365 days, which is the common convention for uptime calculators. Most real SLAs are measured over a calendar month instead, so a 28-day February allows a few minutes less downtime at the same percentage than the figures above. Check which one your contract means before you rely on a number from any calculator, including this one.
How to calculate uptime
Uptime is the share of a period a service was working. The same formula runs in both directions:
Uptime % = (total time - downtime) ÷ total time × 100
Allowed downtime = total time × (1 - uptime % ÷ 100)
A worked example, for 99.9% over a month of 30 days. 99.9% uptime leaves 0.1% of the month for downtime. The month is 2,592,000 seconds long, and 0.1% of that is 2,592 seconds, or 43m 12s. Run it backwards and 2,592 seconds of downtime in 2,592,000 seconds works out to 99.9% uptime.
Availability is calculated the same way, so an availability calculator and an uptime calculator give the same answer for the same target. The difference is emphasis: uptime usually names what a service achieved, and availability the target it is held to. Read more about availability and the nines, or about the SLO a team sets itself.
Knowing the number is the easy half
Hitting it means finding out before your customers do. Uptimely checks as often as every minute, pages the person on call, and publishes the result on a status page — one product, not three. See what it costs.
The Free plan has 10 monitors, with no credit card and no time limit.
Publish the number you are hitting
A status page turns measured uptime into something customers can check themselves. Build a status badge to put the current state in your README, or see what shipped recently.
How do you calculate uptime and downtime?
Subtract the downtime from the total time, divide by the total time, and multiply by 100. Over a month of 30 days, 43m 12s of downtime is 0.1% of the month, so uptime is 99.9%. To go the other way, multiply the total time by the share left for downtime. This calculator does both.
How much downtime is 99.9% uptime?
43m 12s a month, 8h 45m 36s a year, and 1m 26s a day. Those figures count a month as 30 days and a year as 365 days.
How much downtime does 99.5% uptime allow per month?
3h 36m a month, 10h 57m a quarter, 1d 19h 48m a year, and 7m 12s a day. A month here is 30 days, the common convention.
Is an availability calculator the same as an uptime calculator?
For a target measured in time, yes. Availability and uptime are the same ratio, the share of a period a service was usable, so 99.95% allows 21m 36s of downtime a month whichever word the contract uses. The usual difference is emphasis: uptime tends to mean what a service achieved, and availability the target it was held to.
How much downtime does 99.99% uptime allow?
4m 19s a month and 52m 34s a year. Going from 99.9% to 99.99% takes the monthly budget from 43m 12s down to 4m 19s. The engineering cost of that step is usually where the real argument is.
What does five nines mean?
Five nines is 99.999% uptime: 5m 15s of downtime a year, or 26s a month. It is a level almost nothing outside core infrastructure is genuinely built to, and promising it in a contract commits you to responding faster than most teams can even be paged.
Is 99.9% uptime good?
For most software businesses it is a reasonable target. Know what you are agreeing to before you sign it: 43m 12s a month is one bad deploy, and a single incident that runs past lunch can spend a whole quarter's budget.
How long is a month in these calculations?
30 days, a quarter is 91.25 days (365 ÷ 4), and a year is 365 days. That is the common 30-day-month convention, so figures are comparable with other calculators that use it. Real SLAs are usually measured over a calendar month, which runs from 28 to 31 days. A February allows a few minutes less downtime at the same percentage than the figures here.
Does a short blip count against uptime?
In Uptimely, yes. A failed check is recorded as downtime and shows on your status page even when it does not page anyone. By default a monitor waits for two failed checks in a row before it declares an incident, so a single blip costs you uptime without costing you sleep. The two are deliberately separate decisions.
How do I convert server uptime in seconds to days?
Divide the seconds by 86,400, the number of seconds in a day. A reading of 49,000,000 seconds is 567.1296 days, which is 567 days, 3 hours, 6 minutes, 40 seconds. For hours divide by 3,600 instead. The converter above also reads a days:hours:minutes:seconds time and the line the Linux uptime command prints, so you can paste a reading as the machine gave it. Windows and SNMP counters often count milliseconds or hundredths of a second, so bring those down to seconds first.
How do I check uptime on Windows?
Open Task Manager, go to Performance, then CPU: Up time is shown as days:hours:minutes:seconds, such as 3:04:05:06. In PowerShell 7, Get-Uptime returns the same span, and (Get-Uptime).ToString() prints it as 3.04:05:06 with a fraction of a second after it. Paste either into the converter above, label and all, and it reads it, as it does the System Up Time line older versions of systeminfo printed.
How do I measure my actual uptime?
Point a monitor at the thing your customers use and let it check on a schedule. Uptimely records whether each check passed and how long it took, so uptime is measured rather than estimated. A status page can show the uptime percentage for each component it lists.