99.99% uptime: how much downtime is that?

99.99% uptime in exact figures

Most SLAs are measured per month, and a real month is 28 to 31 days long. The same 99.99% allows this much in calendar months and years:

Formula: downtime = (100 - 99.99) / 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 99.99% uptime means in practice

Typical for: Core cloud infrastructure such as multi-zone compute and load balancers, payment and login services, and platforms other companies run their business on.

• People are too slow. By the time someone is paged and logged in, most of the month's budget is gone. Failover, restarts and rollbacks have to happen without a person.
• One zone is not enough. Amazon EC2 and Google Compute Engine only commit to 99.99% for VMs running in more than one zone.
• Every dependency counts. Three services at 99.99% each in a chain give about 99.97% together, so each part has to be better than the whole.

Maintenance windows

There is no room for planned downtime. Maintenance happens with the service up: rolling restarts, online schema changes, and traffic moved away from whatever is being worked on. One 30-minute window counted as downtime uses 684% of the monthly budget, more than the whole month allows.

Monitoring and alerting

Check every 30 seconds or faster, from more than one location, and send alerts to automation as well as to people. A 5-minute check interval is longer than the whole monthly budget.

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 4m 23s a month that 99.99% 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 99.99% budget

A zone fails and the load balancer routes around it

At four nines the timeline has to run without a person. Here one availability zone stops serving and the load balancer moves traffic to the other zones on its own.

Users saw errors for 45s, 17% of the 4m 23s budget, which leaves 3m 38s. The month has room for five failovers like this; the sixth breaches the SLA. If a person had to make the switch, the 5 minutes it takes to answer a page would already be more than the whole month allows.

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 99.99%

• Amazon EC2 commits to 99.99% monthly uptime in a region for instances running in more than one availability zone. A single instance gets 99.5%. Amazon Compute SLA
• Google Compute Engine commits to 99.99% monthly uptime for instances in multiple zones and for load balancing, in most regions (Premium Tier network). Compute Engine SLA

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

99.99% 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 99.99% uptime?

About 4.38 minutes a month (4m 23s) and 52.59 minutes a year (52m 35.7s). Per day it is 8.6s.

Is 99.99% uptime realistic?

Yes, for systems built for it: redundant at every layer, deployed without downtime, and able to recover on their own. For a single server or a single database it is not, because one restart or one slow failover can take more than 4m 23s.

How much downtime is four nines per year?

52m 35.7s in an average year of 365.2425 days, which is about 52.59 minutes.