99.9% uptime: how much downtime is that?

99.9% uptime in exact figures

Most SLAs are measured per month, and a real month is 28 to 31 days long. The same 99.9% allows this much in calendar months and years:

Formula: downtime = (100 - 99.9) / 100 × period. The averages use a 365.2425-day year (the average Gregorian year, leap days included), a month of 1/12 of that and a quarter of 1/4. How the calculator works

What 99.9% uptime means in practice

Typical for: Web apps and APIs on a single primary database, SaaS plans without a premium support tier, internal tools, and single cloud VMs.

• One real incident a month fits. A 40-minute outage stays inside the monthly budget, a second one in the same month does not.
• A person can still fix it. There is time to get paged, open a laptop and roll back a bad deploy by hand, as long as that starts within minutes.
• One server or one zone can be enough, if restarts and deploys are quick. You do not need multi-region failover, but you do need deploys that do not take the site down.

Maintenance windows

If your SLA counts planned maintenance as downtime, keep windows short and rare. Many contracts exclude maintenance that is announced in advance, so check the exclusions in yours. One 30-minute window counted as downtime uses 68% of the monthly budget.

Monitoring and alerting

Check at least every minute. A 5-minute check interval can miss a short outage completely, and lets a long one run for up to 5 minutes before anyone knows.

An outage that starts just after a check can run for a whole interval before the next check sees it. This is what one such interval costs out of the 43m 49.7s a month that 99.9% allows:

Drumbeats uptime checks run as often as every 60 seconds on the Free plan and every 30 seconds on paid plans; you set the interval for each monitor. A failed check is retried twice, 5 seconds apart, before it counts as down, so a single failed request does not page you. Uptime monitoring

How one incident spends the 99.9% budget

A bad deploy, rolled back by hand

A typical three-nines incident: a deploy breaks checkout on an app that runs on one server, and a person has to notice and roll it back. The uptime check runs every minute.

Checkout was down for 16m 10s, which uses 37% of the 43m 49.7s a month allows and leaves 27m 39.7s. The month has room for two incidents like this; the third breaches the SLA. The fix took longest, so quick rollbacks are what keep three nines safe.

Downtime counts from the moment the service fails, not from when a check or a person notices it, so the time it takes to detect a failure comes out of the budget as well as the time to fix it.

Who commits to 99.9%

• Google Compute Engine commits to 99.9% monthly uptime for a single VM of most machine families (Premium Tier network). Compute Engine SLA
• Amazon S3 Standard pays service credits when monthly uptime falls below 99.9%. Amazon S3 SLA

Checked against each provider's SLA page in October 2026. Providers change their terms; the linked page is the one that counts.

99.9% compared with nearby levels

Each extra nine cuts the downtime budget by a factor of 10. Full table from 90% to 99.999%

Questions

How many minutes of downtime is 99.9% uptime?

About 43.83 minutes a month (43m 49.7s) and 8.77 hours a year (8h 45m 57s). Per day it is 1m 26.4s.

Is 99.9% uptime per month or per year?

It depends on the window in the SLA. Cloud SLAs usually measure each month on its own, so the 43m 49.7s cannot be saved up: a quiet January does not cover a bad February. A yearly window allows 8h 45m 57s, which a single long outage can use up.

Is 99.9% uptime good?

For most web apps, yes. It leaves room for one real incident a month and does not need multi-zone infrastructure. If an hour of downtime costs you customers or money, aim for 99.95% or 99.99% and budget for the redundancy they need.