SLA Downtime Calculator

Why This Calculator Exists

"We guarantee 99.99% uptime" sounds harmless in a sales call. It means four and a half minutes of total downtime per month — less than one bad deploy. Plenty of teams sign SLAs without ever converting the percentage into actual minutes, and the gap between what the contract says and what the architecture can deliver only shows up when a customer asks for service credits. This tool does the conversion before you sign, not after.

How to Use It

Type any uptime percentage into the input — it takes up to four decimal places, so 99.95 or 99.9999 both work. The allowed downtime per day, week, month, and year updates as you type. If your number matches one of the standard "nines", that row lights up in the comparison table below so you can see how your commitment stacks against the usual tiers. Copy as Markdown puts the whole table on your clipboard, formatted and ready to paste into an RFC, a runbook, or an architecture doc.

How It's Calculated

downtime = (1 − uptime%) × period_in_minutes
Periods used: 1440 min/day · 10080 min/week · 43800 min/month · 525960 min/year.

The "Nines"

• 99% — "two nines" — ~7.3 h/month. Consumer grade.
• 99.9% — "three nines" — ~43.8 min/month. Typical SaaS commitment.
• 99.95% — ~21.9 min/month. Common for managed cloud services.
• 99.99% — "four nines" — ~4.4 min/month. High-availability target.
• 99.999% — "five nines" — ~26 sec/month. Carrier-grade.

Frequently Asked Questions

Why is a month 43,800 minutes?

That's the average month: 365.25 days divided by 12, or 30.4375 days. Some SLAs measure against the actual calendar month instead, which shifts the numbers slightly — a 28-day February is more forgiving than a 31-day July at the same percentage.

Does this account for planned maintenance?

No. Most SLA contracts exclude scheduled maintenance windows from the downtime budget, but not all of them do. The calculator gives you the raw math; the contract language decides what counts against it.

What's a realistic uptime to commit to?

A single-region deployment with a solid on-call rotation can usually hold 99.9%. Beyond that, each extra nine tends to demand structural changes — redundancy, automated failover, multi-region — not just more careful operations. Committing to four nines with a single database is a promise your architecture can't keep.

Is downtime measured per incident or in total?

In total, across the measurement period. Three ten-minute outages in a month blow through a 99.95% budget just as thoroughly as one thirty-minute outage does.