SLA & Downtime Calculator

An SLA number like 99.9% sounds indistinguishable from perfect. It isn't. Availability is just (total time − downtime) ÷ total time, so every SLA percentage is really a downtime budget in disguise: 99.9% means your provider may be down for 43 minutes 50 seconds every month — almost 9 hours a year — and still owe you nothing. The whole point of this calculator is to force that conversion, because nobody's intuition works in percentages of a month.

The "nines" ladder is logarithmic: each extra nine cuts the budget by a factor of ten, and each step changes how you have to operate. At 99% (7+ hours a month) a human can notice an outage, have coffee, and fix it. At 99.9% you need alerting that pages someone within minutes. At 99.99% — 4 minutes 23 seconds a month — a human can no longer be in the loop at all; recovery has to be automated, because by the time someone has acknowledged the page the budget is gone. And 99.999% allows 26 seconds a month, less time than it takes to SSH into a box. That is why the jump from 99.9% to 99.99% is not "0.09% better" — it is an entirely different engineering and on-call discipline, usually an order of magnitude more expensive.

Pay attention to the measurement window. Vendors overwhelmingly measure SLAs per calendar month, and it is not an accident: a single 8-hour incident blows a monthly 99.9% SLA spectacularly (that month lands near 98.9%), but the service credit you receive is a slice of one month's bill — and next month the slate is wiped clean. Measured yearly, the same incident would sit inside the annual budget with room to spare. Monthly windows also mean a vendor can fail their SLA twelve times a year and never miss it "annually". When you compare providers, compare the window, not just the number.

Also read the fine print on planned maintenance. Most SLAs carve out announced maintenance windows entirely — that downtime simply doesn't count — and many add further exclusions: incidents shorter than a few minutes, single-region failures, problems "outside the provider's control". A 99.95% SLA with generous carve-outs can be weaker in practice than an honest 99.9%. If you're negotiating your own SLA, the metrics behind it matter as much as the number; our guide to MTTR, MTTD and MTBF explains the recovery-time side of that equation.

Finally: an SLA you don't measure yourself is a number on a sales page. Providers self-report, exclusions apply, and your users experience your stack, not your vendor's. Independent measurement is step one in our uptime monitoring best practices — or start monitoring your own availability free and know your real number instead of trusting theirs.