StatOSS

Uptime calculator

Type in an uptime percentage and see the allowed downtime, or type in the minutes and see the percentage.

%

Allowed downtime:

per day
1m 26s
per week
10m 5s
per month
43m 50s
per year
8h 45m
min

Uptime that leaves:

in a day
97.9167%
in a week
99.7024%
in a month
99.9316%
in a year
99.9943%

The nines

Allowed downtime at each common target. A month here is a twelfth of a year, 30.44 days, which is how most SLAs count.

UptimePer dayPer weekPer monthPer year
90%2h 24m16h 48m3d 1h 3m36d 12h 36m
95%1h 12m8h 24m1d 12h 31m18d 6h 18m
99%14m 24s1h 40m7h 18m3d 15h 39m
99.5%7m 12s50m 24s3h 39m1d 19h 49m
99.9%1m 26s10m 5s43m 50s8h 45m
99.95%43s5m 2s21m 55s4h 22m
99.99%9s1m4m 23s52m 36s
99.995%4s30s2m 11s26m 18s
99.999%1s6s26s5m 16s

Reading the numbers

Two nines is about a quarter of an hour a day. Three nines, the figure most SLAs promise, is under nine hours a year and three quarters of an hour a month. Four nines is under an hour a year, too little for an outage that a person has to notice and fix. Five nines is five minutes a year, which needs automatic failover.

The check interval eats into the budget before anyone knows there is an outage. At a check every five minutes, an outage is found up to five minutes in; at 99.9% a month that is a ninth of the budget spent on noticing. A check a minute, with two failures in a row before it counts, finds the outage within two or three minutes. The uptime monitoring guide goes through the rest of the arithmetic.

The error budget on a page

A StatOSS page with an uptime target draws the month's error budget under the headline: how much downtime the target allows, how much has been spent, and how much is left, from the checks themselves rather than from a figure typed in. Maintenance windows are kept out of it, and the page docs explain how a partial outage counts.