Uptime reference

99.95% uptime

Ninety-nine point nine five percent halves the three-nines budget to about 22 minutes a month. It is the usual price of moving from "we have redundancy" to "we have tested redundancy".

21m 36s of downtime allowed per month

99.95% uptime by period

A month is taken as 30 days, a quarter as 90 days and a year as 365 days.

Period	Allowed downtime	Required uptime
Per day	43s	23h 59m
Per week	5m 2s	6d 23h
Per month	21m 36s	29d 23h
Per quarter	1h 4m	89d 22h
Per year	4h 22m	364d 19h

Where 99.95% is used

Common on paid infrastructure tiers, managed Kubernetes, and the higher tier of a tiered SaaS SLA. Often the level offered to enterprise customers who asked for four nines and negotiated.

What it demands

Automated failover, because 22 minutes is not enough time to wake someone up and have them think. Detection has to be automatic and the recovery path has to already exist.

One step either way

Dropping to 99.9% widens the monthly budget to 43m 12s. Moving up to 99.99% cuts it to 4m 19s.

Measuring it

You cannot report on a service level you are not measuring at a finer resolution than the budget itself. At 99.95% the monthly allowance is 21m 36s, so a check interval anywhere near that figure will miss breaches entirely. Work out your own numbers, or monitor it continuously from four regions, free to start.

Other service levels

Compare the tiers

Every common service level, and the downtime each one allows per month.

90% 3d / month one nine 95% 1d 12h / month 99% 7h 12m / month two nines 99.5% 3h 36m / month 99.9% 43m 12s / month three nines 99.99% 4m 19s / month four nines 99.999% 26s / month five nines

Calculators: Uptime calculator · SLA calculator · Downtime calculator · Availability calculator · Downtime cost calculator

Start free. No credit card.

Uptime, cron, synthetic, logs, RUM, incidents, and status pages. Free tier on every product.

Start free Talk to us