OneWebDesk

Uptime SLA Calculator

Compute allowed downtime per period for 99%–99.999% SLAs.

What does an SLA like “99.9% availability” actually allow in downtime? Enter a target availability (%) and this calculator converts it into the maximum allowed downtime per day, week, month, quarter and year.

Use it to design SLAs, compare cloud/hosting guarantees, or judge whether an outage breached a contract. To also gauge the infrastructure cost as traffic grows, pair it with the bandwidth cost calculator.

Allowed downtime at 99.9% availability
Day1m 26s
Week10m 5s
Month (30d)43m 12s
Quarter (90d)2h 9m 36s
Year (365d)8h 45m 36s

The meaning of "nines"

Availability is often counted in "nines." Each extra nine cuts allowed downtime by roughly 10×.

  • 99%: ~3.65 days/year
  • 99.9%: ~8.76 hours/year
  • 99.99%: ~52.6 minutes/year
  • 99.999%: ~5.26 minutes/year

Allowed-downtime reference table

Per-year figures are hard to feel. Because SLA breaches are usually judged per month (commonly a fixed 30 days), here are the headline availability tiers laid out on the same basis this calculator uses (month = 30 days, quarter = 90 days, year = 365 days).

AvailabilityDayMonth (30d)Year (365d)
90% (one nine)2h 24m3 days36.5 days
95%1h 12m1.5 days18.25 days
99% (two nines)14m 24s7h 12m3.65 days
99.9% (three nines)1m 26s43m 12s8h 46m
99.95%43s21m 36s4h 23m
99.99% (four nines)8.6s4m 19s52m 36s
99.999% (five nines)0.86s26s5m 16s

99.95% and 99.99% are included because they are the tiers most cloud providers publish for standard SLAs (e.g. virtual machines and managed databases).

Worked example: 99.95%

Enter 99.95 and the "downtime fraction" is (100 − 99.95) ÷ 100 = 0.0005, i.e. 0.05%. Multiply that by the total minutes in each window.

  • Month (30d) = 43,200 min × 0.0005 = 21.6 min → 21m 36s
  • Year (365d) = 525,600 min × 0.0005 = 262.8 min → 4h 22m 48s

So a service promising 99.95% that is down for more than ~22 minutes in a single month has breached that month's SLA — typically triggering a service credit worth a set percentage of the bill.

Common pitfall

The most frequent mistake is reading downtime only on a yearly basis. Seeing "99.9% = 8.76 hours/year" makes a day of outage feel acceptable, but most SLAs reconcile per month, where the budget is just ~43 minutes — and a single long incident can burn the entire month's allowance at once. A second trap is averaging nines into a "99.9% on average" claim: availability is judged separately each billing period, not averaged, and serial dependencies multiply it down rather than up.

Frequently asked questions

Is scheduled maintenance counted as downtime?
It depends on the contract. Pre-announced maintenance is often excluded, so check the 'scheduled maintenance' clause.
How many days is a month here?
This tool uses 30 days. Real SLAs may use calendar months or a fixed 30-day basis.
What happens if the SLA is missed?
Most providers offer service credits. These are often claim-based rather than automatic, so check the terms.
How does availability combine when components are in series?
For serial dependencies (DNS to load balancer to app to database), you multiply each component's availability. Three 99.9% components in series yield about 99.7% overall, which is why critical paths are made redundant to raise availability.
Are my inputs sent to a server?
No. All calculations run entirely in your browser, and the target availability you enter is never transmitted or stored anywhere.

Related tools