How much downtime is 99.9% uptime?
Published 19 September 2026
An uptime percentage is easier to reason about once it is turned into time. The table does that for the targets people usually quote. A month here is 30 days and a year is 365.
| Uptime | Per day | Per 30-day month | Per year |
|---|---|---|---|
| 99% | 14m 24s | 7h 12m | 3d 15h 36m |
| 99.5% | 7m 12s | 3h 36m | 1d 19h 48m |
| 99.9% | 1m 26s | 43m 12s | 8h 45m 36s |
| 99.95% | 43s | 21m 36s | 4h 22m 48s |
| 99.99% | 8.6s | 4m 19s | 52m 34s |
| 99.999% | 0.9s | 26s | 5m 15s |
What each extra nine costs
Every extra nine divides the allowance by ten. At 99.9% a service may be down for 43m 12s in a month, which one bad deploy can use up by itself. At 99.99% the monthly allowance is 4m 19s, less time than it takes most teams to notice something is wrong, let alone fix it.
The daily column is the one to be careful with. 99.99% allows 8.6s a day, and no check that runs once a minute can see an outage that short. Targets past three nines only mean something measured over a month or longer.
Measured by whom, and how
A percentage is only as honest as the checks behind it. Before agreeing to one, it is worth settling what counts as down, how often the service is checked, and whether announced maintenance counts against it.
On an InBrief status page, a monitor is marked down after two failed checks in a row, so a single dropped packet does not publish an outage, and one successful check brings it back up. Checks that fall inside a scheduled maintenance window are left out of uptime, so work you announced in advance does not count against the figure your customers see.