Five Nines (99.999%) Uptime Calculator
Five nines means 99.999% availability: a maximum of 26 seconds of downtime per 30-day month and 5 minutes 15 seconds per year. At that level no human can participate in recovery, and every component in the critical path must be redundant across regions — which is why five nines is common in telecoms and rare in ordinary web applications.
Per day
1.4 min
Per week
10.1 min
Per month (30d)
43.2 min
Per quarter
2.2 h
Per year
8.8 h
Join the Veojson dispatch
Get the SRE reliability starter kit with SLO templates and alerting thresholds.
One email, no spam, unsubscribe any time.
What five nines actually demands
Active-active deployment across independent regions, with traffic shifted automatically by health signal rather than a runbook. Data replicated synchronously or with a tested, automatic promotion path. Deployments that are progressive and automatically reverted on error-rate regression. No shared single point of failure — including the DNS provider, the certificate authority, the CI pipeline that can push a bad config everywhere at once, and the observability stack you need in order to notice.
The cost curve
Each nine typically multiplies infrastructure and engineering cost several times over while reducing downtime tenfold. Going from 99.9% to 99.99% removes 39 minutes of monthly downtime; going from 99.99% to 99.999% removes fewer than 4 more. Before committing, ask what a minute of downtime genuinely costs the business — for most products the honest answer makes 99.95% or 99.99% the rational target, and the money is better spent on faster recovery than on another nine.
Five nines and correlated failure
Redundancy only helps when failures are independent. Two replicas at 99% give 99.99% combined — but only if they do not share a region, a deployment, a certificate, a configuration push or a dependency. Most real outages at high-availability shops are correlated: a bad config rolled out everywhere, an expired certificate on every node, a full disk on every replica at once. Chaos testing and staggered rollouts matter more than adding another replica.
Five nines (99.999%) allowed downtime
| Period | Allowed downtime |
|---|---|
| Per day | 0.9s |
| Per week | 6s |
| Per 30-day month | 26s |
| Per quarter | 1m 18s |
| Per year | 5m 15s |
The nines, side by side
| Nines | Availability | Per month | Per year |
|---|---|---|---|
| Two | 99% | 7h 12m | 3d 15h 36m |
| Three | 99.9% | 43m 50s | 8h 45m 57s |
| Four | 99.99% | 4m 23s | 52m 36s |
| Five | 99.999% | 26s | 5m 15s |
| Six | 99.9999% | 2.6s | 31.6s |