Skip to content

Uptime & SLA Calculator

Turn an SLA percentage into the downtime it allows, work out uptime from real downtime, combine services in series and in parallel, and set alerts on how fast an error budget burns.

Free · No signupWorked out in your browser

The availability the SLA or SLO promises.

A month is
A year is

Match your contract: many bill by the calendar month, which runs 28 to 31 days.

Allowed downtime for

99.9%

three nines

Allowed downtime for each period
PerAllowed downtimePeriod
Day1m 26.4s1 day
Week10m 4.8s7 days
Month43m 12s30 days
Quarter2h 11m 24s91.25 days
Year8h 45m 36s365 days

That is your error budget. A 5-minute outage spends 11.6% of the month's, and one deploy with 30 seconds of errors spends 1.16%.

Worked out in this tab as you type. Nothing you enter leaves this page.

Budgets are theory: know your real numbers

Watch each server’s health, live.

CtrlOps gives you real-time CPU, memory and disk for each server you open, and alerts such as Website down, so you know how much of your error budget you are actually spending. Local-first, credentials never leave your machine.

See server monitoringDownload for MacDownload for Windows

The nines

Allowed downtime,
by SLA.

With a 365-day year and a 30-day month. Each extra nine is ten times less room.

Allowed downtime for common SLAs
SLANinesPer yearPer monthPer weekPer day
99%Two3d 15h 36m7h 12m1h 40m 48s14m 24s
99.5%Two and a half1d 19h 48m3h 36m50m 24s7m 12s
99.9%Three8h 45m 36s43m 12s10m 4.8s1m 26.4s
99.95%Three and a half4h 22m 48s21m 36s5m 2.4s43.2s
99.99%Four52m 33.6s4m 19.2s1m 0.5s8.64s
99.999%Five5m 15.4s25.9s6.05s0.864s

How uptime is counted

A percentage.
A number of minutes.

What uptime, SLA, and the nines mean

Uptime is the share of time a service is available. An SLA, a service level agreement, promises a minimum, usually as a percentage like 99.9%. People count the leading nines: 99.9% is three nines, 99.999% is five. Each extra nine cuts the allowed downtime by ten.

How allowed downtime is calculated

It is the length of the period times the share you are not promising: downtime = period × (1 − SLA / 100). For 99.9% over a 30-day month that is 2,592,000 s × 0.001 = 2,592 s, or 43m 12s. The window matters, which is why the calculator lets you set how long a month and a year are.

Error budgets and burn rate

Site reliability teams turn the SLA around: the allowed downtime is an error budget to spend on incidents, risky deploys and maintenance. The burn rate is how fast it is going: at 1×, a 30-day budget lasts exactly 30 days; at 14.4×, it is gone in two days, and an hour of that spends 2% of the month. That is the threshold Google's SRE workbook pages on, checked over a long and a short window so a blip does not wake anyone.

Several services: series and redundancy

A request that needs every one of several services is up only when all of them are, so availabilities multiply: three at 99.9% give about 99.7%. Redundant copies work the other way: two copies at 99.9% are both down only 0.0001% of the time, if they fail independently. Shared power, a shared network or the same bad deploy breaks that assumption, so treat the result as a ceiling.

SLA, SLO and SLI

  • The SLI is what you measure, such as the share of requests that succeed.
  • The SLO is your own target for it, such as 99.95%.
  • The SLA is the promise to customers, usually looser than the SLO and often with penalties.

Gotchas

  • Monthly and yearly SLAs allow very different downtime; check which your provider uses.
  • Many SLAs leave planned maintenance out of the count.
  • Partial outages, raised error rates and slow responses may or may not count: the SLI decides.
Get alerted before the budget is gone

Uptime & SLA FAQ

Uptime and SLAs,
explained.

Keep reading

Tools and guides for your next step.

Your privacy preferences. We use product analytics to improve CtrlOps. Google measurement runs without analytics cookies. Live chat stays off until you allow it. Cookie policy