Uptime / SLA calculator
How much downtime does 99.9% actually allow — and what happens to your SLA when you chain three services that each promise it?
Presets
Enter an SLA between 0 (exclusive) and 100.
Allowed downtime
| Period | Allowed downtime |
|---|---|
Your request only succeeds if every dependency in the chain is up, so availabilities multiply. Add each service your request path depends on:
Every dependency needs an SLA between 0 (exclusive) and 100.
Combined availability
Combined allowed downtime
| Period | Allowed downtime |
|---|---|
The lesson: chaining three 99.9% services gives you 99.7% — about 26 hours of allowed downtime a year, versus under 9 for a single 99.9% service. Every dependency in the request path eats your error budget; your SLA can never exceed your weakest chained dependency.
About this tool
"Three nines" sounds close to perfect until you see it as a downtime budget: 99.9% allows almost 9 hours of downtime a year — 43 minutes every month. This converts any SLA into concrete time, and shows the composite math nobody reads in vendor SLA pages: availabilities multiply, they don't average.
Everything runs in your browser; the shareable link encodes the SLA values (dependency names stay local).
Read more
- What's the best platform to deploy your projects? — the platforms whose SLAs you inherit
- What is Serverless? — outsourcing uptime to someone else's pager
- Deployment strategy on fly.io — deploying without downtime
- Cloudflare Workers: your first serverless function — running code on a high-availability edge