Uptime / Downtime Calculator

Enter a target uptime percentage and see what it actually allows in downtime — per day, week, month, and year. Add your revenue to estimate cost. Toggle compare mode to see what moving from one SLA tier to another buys you.

%
$
Downtime per year
at 99.9% uptime
Period Downtime allowed Estimated cost
Per week
Per month
Per quarter
Per year

How to use this calculator

Enter your target uptime percentage in the top field — or click one of the preset SLA buttons (98 / 99 / 99.9 / 99.95 / 99.99). The calculator immediately shows what that target allows in real downtime per day, week, month, and year.

If you want dollar numbers on it, fill in the revenue field (yearly or monthly). The tool assumes downtime causes proportional revenue loss — a useful first approximation but a conservative one. Real costs are usually 2–5x what this number shows once you include incident response, churn, and reputational damage.

Toggle Compare two SLAs side-by-side to see what moving from one tier to another buys you in absolute terms — useful when the engineering team is debating whether 99.99 is worth the extra resilience work over 99.9.

Common SLA benchmarks

What each common uptime tier translates to in actual downtime. The cost curve gets exponentially harder with each additional nine.

Uptime Annual downtime Monthly downtime Daily downtime Typical context
90%36.5 days3 days2h 24mHobby project
98%7.3 days14h 24m28m 48sInternal tools
99%3.65 days7h 12m14m 24sB2B SaaS, low tier
99.9% (three nines)8h 45m43m 12s1m 26sStandard consumer SaaS
99.95%4h 22m21m 36s43.2sPremium SaaS, e-commerce
99.99% (four nines)52m 33s4m 19s8.6sBanking, healthcare
99.999% (five nines)5m 15s26s0.86sTelecom, life-safety

What 99.9% uptime means in real terms

Three nines (99.9% uptime) sounds like a high standard until you do the math. Over the course of a year, a 99.9% target gives you almost 9 hours of allowed downtime — that's a full working day of being offline, distributed across 12 months. Whether that's acceptable depends on what you're running.

If you're running a global SaaS, 9 hours of unannounced downtime is the difference between "we had a tough quarter" and "we lost the enterprise contract." If you're running a hobby project, it's probably fine.

Most consumer-grade web services land in the 99.9–99.95 range. Banking and healthcare systems aim for 99.99. Telecoms and life-safety systems aim for 99.999. The cost curve is exponential — adding each additional nine roughly doubles the engineering investment. Most teams over-promise on availability they're not actually hitting.

Two things to keep in mind that the calculator does not capture:

Planned maintenance counts as downtime — usually. Most internal SLOs exclude scheduled maintenance windows; most external SLAs include them. If your SLO says 99.9% but you're allowed to schedule 4 hours of maintenance per month outside of it, your effective customer experience is closer to 99.5%. Read the fine print on whose definition you're using.

Multi-region and multi-service compounds. If your application depends on three independent services each at 99.9%, your application's effective uptime is 99.9% × 99.9% × 99.9% = 99.7%. The more dependencies you stack, the lower the practical ceiling — which is why "pick four nines or higher" sounds easy and is in practice extremely hard.

For a deeper take on choosing the right target and turning it into a measurable SLI, read our practical guide to SLA, SLO, and SLI.

FAQ

What's the difference between uptime and availability?
In practice, nothing — both refer to the percentage of time a service is responding correctly. "Availability" is more common in formal SLAs; "uptime" is more common in marketing copy.
How does this calculator estimate cost?
It assumes downtime causes a proportional revenue loss: if you're down 0.1% of the time, you lose 0.1% of revenue. This is a useful first approximation but ignores second-order effects like customer churn, lost trust, and engineering time spent on incident response. The real cost of downtime is usually 2–5x what this number shows.
Why is 99.99% (four nines) so much harder than 99.9% (three nines)?
Three nines gives you ~9 hours of error budget per year. Four nines gives you ~52 minutes. At three nines, you can absorb a single bad incident and still hit your target; at four nines, one bad afternoon blows your annual budget. The engineering practices required (hot standby, automated failover, multi-region) are categorically different.
Should my SLO equal my SLA?
No. Your SLO (internal target) should be tighter than your SLA (external commitment). If you commit to 99.9 externally and target 99.9 internally, you have zero margin for error. Most teams set the SLO one nine higher than the SLA — internal target 99.99, external commitment 99.9.
How do I actually measure uptime for my service?
You need a continuous external monitor (HTTP, ICMP, or synthetic) that reports availability over the period you care about. Percentage uptime = successful_checks / total_checks. For a deeper take on choosing the right SLI denominator, see our SLA / SLO / SLI guide.
Why are the per-month and per-year numbers slightly different from other calculators?
We use the conventional 365-day year (525,600 minutes) and 30-day month (43,200 minutes). Some calculators use 365.25 days for the year (factoring in leap years), which produces marginally different numbers. The difference is well within rounding for any practical SLO discussion.

Related from the Oack blog

Want this monitored continuously?

Oack has a free tier for the first 5 monitors. TCP-level diagnostics, 60-second checks, and instant Slack alerts when uptime drops.