How much downtime does 99.5% uptime allow?

99.5% uptime allows about 3h 36m of downtime per 30-day month, which works out to roughly 1d 20h a year. Go past that in a window and you have missed the target for that window.

What is the error budget for 99.5% uptime?

Over 5M requests in a month, 99.5% permits about 25K failed requests before the budget is spent. The budget is 0.5% of whatever volume you serve, so it scales with traffic.

99.5% uptime downtime & error budget

99.5% uptime allows about 3h 36m of downtime per 30-day month, roughly 1d 20h a year. Here is the full ladder, the error budget in requests, and how fast you can burn it.

allowed downtime / month3h 36m30-day window

error budget25K reqsof 5M / month

failure allowance0.5%of all requests

allowed_downtime

Allowed downtime at 99.5%, by window

Allowed downtime at 99.5% across time windows
window	allowed downtime
per day	7m 12s
per week	50m 24s
per 30 days	3h 36m
per 90 days	10h 48m
per 365 days	1d 20h

burn_rate

How fast the monthly budget burns

At 1x you spend the whole 30-day budget exactly over 30 days. 14.4x is Google's fast-burn alert threshold, the rate that drains the month in about two days.

Time to exhaust the 30-day budget at each burn rate
burn rate	budget gone in
1x	30d
2x	15d
5x	6d
10x	3d
14.4x (fast-burn page)	2d 2h

free_toolRun your own volumeThe SLO & Error Budget Calculator, prefilled to 99.5%. Set your real request volume to size the budget and the burn-rate alerts.free_toolWhat is that downtime worth?The Downtime Cost Calculator puts a dollar figure on this tier and on each extra nine, so you can decide what reliability is worth buying.

what_it_takes

What it takes to hold 99.5%

A common informal bar for early internal tools and B2B software with patient users. You reach it with health checks, automated restarts, and a deploy process that rolls back fast. It still tolerates the occasional manual recovery, which is why it is cheap to hold and rarely good enough to put in a contract.

Need more? See 99.9% uptime and what the next nine costs. Lighter need? 99% uptime is cheaper to hold.

every_tier

The full uptime ladder

New to the terms? What availability and the nines mean, plus error budgets and burn rate.

faq

Questions & answers

How much downtime does 99.5% uptime allow?: 99.5% uptime allows about 3h 36m of downtime per 30-day month, which works out to roughly 1d 20h a year. Go past that in a window and you have missed the target for that window.
What is the error budget for 99.5% uptime?: Over 5M requests in a month, 99.5% permits about 25K failed requests before the budget is spent. The budget is 0.5% of whatever volume you serve, so it scales with traffic.
Is 99.5% uptime good enough?: It depends on what your users need and what the next tier costs to hold. A common informal bar for early internal tools and B2B software with patient users. The breakdown below shows what it takes to hold 99.5% and whether the next nine is worth it.

Picking a target is easy. Holding it in production is the work.

I review where a system actually spends its error budget and what the next nine really costs. Book a call, or leave your email and I'll reach out.

Book a call

Prefer proof first? See how this plays out in real case studies →