How much downtime does 90% uptime allow?
90% (one nine) uptime allows 36d 12h of downtime per year. Here is the full budget, by period.
| Per day | 2h 24m |
| Per week | 16h 48m |
| Per month | 3d |
| Per quarter | 9d |
| Per year | 36d 12h |
Is 90% (one nine) the right target?
One nine is a low bar — it suits internal tools, early prototypes, and non-critical batch jobs where the occasional outage costs nothing. It is far too loose for anything customer-facing: at 90% a service can be unavailable for hours at a stretch without breaching the target.
What it takes to hold 90%
Little more than a single server and basic monitoring. With no redundancy and no on-call coverage, a failure can sit unresolved for hours — which is exactly why 90% is only acceptable for low-stakes workloads.
Where you’ll see 90%
- An internal admin dashboard used only during office hours
- A staging or QA environment
- A nightly batch or reporting job
Each additional nine cuts the allowed downtime by roughly tenfold. The chart below shows where 90% sits relative to the other common uptime targets — and why detection speed, not just the headline percentage, decides whether you actually hold the line.
Frequently asked questions
- How much downtime is 90% uptime?
- 90% uptime allows about 36 days 12 hours of downtime per year — roughly 3 days per month, or 2 hours 24 minutes every single day. That is why it is unsuitable for anything customers depend on.
- Is 90% uptime good enough for internal tools?
- For low-stakes internal tools, prototypes, or off-hours batch jobs it can be acceptable, because a short outage rarely costs money. The moment customers or revenue depend on the service, you should target 99.9% or higher.
- How do I monitor a service at the 90% level?
- Even a slow check interval will catch a 90%-class outage, but faster detection still shortens recovery. PingInsight's free plan checks every 60 seconds, so you learn about failures in about a minute instead of hours.
Compare every tier on the full uptime / SLA calculator, read what “three nines” really means, or see how 1-second uptime monitoring helps you stay inside this budget.
| Period | Allowed downtime |
|---|---|
| Per day | 2h 24m |
| Per week | 16h 48m |
| Per month | 3d |
| Per quarter | 9d |
| Per year | 36d 12h |
Detection matters as much as the budget. With 5-minute monitoring you can burn minutes of this allowance before you even detect an outage. PingInsight's true 1-second checks catch it almost immediately.
Back to the full uptime calculator or see how 1-second checks help you stay within budget.