How much downtime does 99.9% uptime allow per month?

On a 30-day month (2,592,000 seconds), 99.9% uptime allows 2,592 seconds of downtime — that is 43 minutes 12 seconds. Over a full 365-day year it works out to 8 hours 45 minutes 36 seconds, the widely-cited “99.9% ≈ 8.76 hours/year” figure.

What is the difference between 99.9% and 99.99% uptime?

Each extra nine cuts the allowed downtime roughly tenfold. 99.9% (three nines) permits about 43 minutes of downtime a month; 99.99% (four nines) permits only about 4 minutes 19 seconds a month. Over a year that is 8.76 hours versus 52.6 minutes.

How do you calculate uptime percentage from downtime?

Uptime % = (total period − downtime) ÷ total period × 100. For example, 3 hours 40 minutes (13,200 seconds) of downtime in a 30-day month gives (2,592,000 − 13,200) ÷ 2,592,000 × 100 = 99.4907%. This is the same “Monthly Uptime Percentage” formula AWS, Google Cloud and Azure publish.

What does “five nines” availability mean?

Five nines is 99.999% uptime. It allows only about 0.86 seconds of downtime per day, 25.9 seconds per month, and roughly 5 minutes 15 seconds per year. It is the gold standard for telecom and critical infrastructure, and is expensive to engineer because it leaves almost no margin for failure.

How are AWS, Google Cloud and Azure SLA service credits calculated?

Each cloud publishes credit bands tied to your monthly uptime. AWS EC2 and Google Compute Engine commit to 99.99%: below that down to 99.0% earns a 10% credit. Below 99.0% earns 30% (AWS) or 25% (Google), and below 95.0% earns 100%. Azure single-instance VMs commit to 99.9%, with 10%, 25% and 100% bands. The credit is a percentage of the monthly bill, applied as future credit.

Does this calculator use a 30-day month or a calendar month?

The “Per month” figures use a 30-day month (2,592,000 seconds), which is the convention most SLA tables use. You can also pick a 31-day month in the reverse mode. Cloud providers technically measure against the actual minutes in each billing month, so for a 31-day month the allowed downtime is slightly higher.

What if my downtime was longer than the whole period?

The tool clamps downtime to the period length, so achieved uptime never drops below 0% or shows a negative number. If you enter downtime that exceeds the period, it flags this and reports 0% uptime — meaning the service was effectively down for the entire window.

Are the cloud service-credit figures kept up to date?

The AWS, Google Cloud and Azure credit bands are a static snapshot reviewed quarterly, last verified on 2026-06-19. Provider SLAs change rarely and need human verification, so we do not scrape them live. The core downtime math is provider-independent arithmetic and never goes stale.

Developer · Reliability

Uptime / SLA Downtime Calculator

Turn an uptime promise like 99.9% into the exact downtime it allows — per day, week, month and year — or reverse from real downtime to the uptime you achieved. Maps onto AWS, Google Cloud and Azure service-credit bands. No signup, sources cited below.

By Induwara Ashinsana— Executive Director, Ryzera TechnologiesUpdated Jun 19, 2026

Uptime / SLA Calculator

Provider bands verified · 2026-06-19

Uptime target (%)

How much availability you promise. Up to 5 decimals supported.

Common SLAs

Allowed downtime at 99.9%

That’s the “three nines” availability class.

Per day

1m 26.4s

1.44 min

Per week

10m 4.8s

10.1 min

Per month

43m 12s

43.2 min

Per quarter

2h 9m 36s

129.6 min

Per year

8h 45m 36s

525.6 min

The “nines” reference

Uptime	Class	Per month	Per year
90%	one nine	2d 23h 59m 60s	36d 11h 59m 60s
95%	—	1d 12h 0s	18d 6h 0s
99%	two nines	7h 12m 0s	3d 15h 36m 0s
99.5%	—	3h 36m 0s	1d 19h 48m 0s
99.9%you	three nines	43m 12s	8h 45m 36s
99.95%	—	21m 36s	4h 22m 48s
99.99%	four nines	4m 19.2s	52m 33.6s
99.999%	five nines	25.92s	5m 15.36s

“Per month” uses a 30-day month (2,592,000 s); “per year” uses 365 days.

The downtime math is provider-independent arithmetic. Service-credit bands are a static snapshot of AWS, Google Cloud and Azure SLAs, last verified 2026-06-19. See sources below.

How it works

Uptime and downtime are two views of the same number. If a service is available for fraction u of a period, it is unavailable for 1 − u. Multiply that by the length of the period and you get the downtime budget. This is exactly how the major clouds define their Monthly Uptime Percentage: (total minutes − downtime minutes) ÷ total minutes.

Uptime % → allowed downtime. Convert the percentage to a fraction (99.9% → 0.999), subtract from 1 (0.001), and multiply by the period in seconds. A 30-day month is 2,592,000 seconds, so 0.001 × 2,592,000 = 2,592 seconds = 43m 12s.
Downtime → achieved uptime %. Take the observed downtime in seconds, subtract it from the period, divide by the period and multiply by 100. The figure is kept at full precision internally and only rounded for display, so boundary cases classify correctly.
Availability class (the “nines”). The achieved uptime is compared against the standard ladder — 90% (one nine), 99% (two nines), 99.9% (three nines), 99.99% (four nines), 99.999% (five nines), plus the half-steps 95%, 99.5% and 99.95%. The tool reports the highest class met and the next one missed.
Provider service credit.When you pick a provider lens, the achieved uptime is looked up in that provider's published credit schedule. If it sits below the SLA commitment, the matching band's credit percentage (10%, 25/30%, or 100% of the monthly bill) is shown with the exact band cited.

The reverse calculation is cross-checked against the cloud providers' own minute-based formula: 3h 40m of downtime is 220 minutes; in a 30-day month of 43,200 minutes that is (43,200 − 220) ÷ 43,200 × 100 = 99.4907%, which matches the seconds-based result to the last decimal. Both paths are pure arithmetic with no rounding before the final display step, so a value sitting exactly on a tier boundary (say 99.99%) classifies as that tier and not the one below it.

Worked examples

Three nines

99.9% over a 30-day month

Period: 30 days = 2,592,000 seconds
Downtime fraction: 1 − 0.999 = 0.001
Allowed downtime: 2,592,000 × 0.001 = 2,592 s
Formatted: 43m 12s per month
Per year: 31,536,000 × 0.001 = 31,536 s = 8h 45m 36s

Reverse + credit

3h 40m down in a month, Azure lens

Downtime: 3×3600 + 40×60 = 13,200 s
Period: 2,592,000 s (30-day month)
Uptime: (2,592,000 − 13,200) / 2,592,000 × 100 = 99.4907%
Class: clears 99% (two nines), misses 99.5%
Azure VM band “< 99.9% and ≥ 99%” → 10% service credit

Edge case

Five nines over a single day

Target: 99.999% uptime
Period: 1 day = 86,400 seconds
Downtime fraction: 1 − 0.99999 = 0.00001
Allowed downtime: 86,400 × 0.00001 = 0.864 s ≈ 864 ms
Per year: 31,536,000 × 0.00001 = 315.36 s ≈ 5m 15s

Frequently asked questions

Sources & references

The downtime math is provider-independent and never goes stale. The AWS, Google Cloud and Azure service-credit bands are a static snapshot last cross-checked on 2026-06-19and reviewed quarterly. Email me if a provider has changed its bands and I'll update them.

Related tools

LiveUtility

Body Fat %

Estimate your body-fat percentage from neck, waist, and hip tape measurements using the original US Navy (Hodgdon & Beckett 1984) method. Metric and imperial, men and women, with lean body mass and ACE fitness bands — runs entirely in your browser, no signup.

Open tool

LiveUtility

Speech Time Calculator

Convert words to spoken minutes and back at slow, conversational, average or fast pace. Find how many words a 5-minute speech needs, compare paces, and see silent-reading time. Free, instant, runs in your browser.

Open tool

LiveUtility

Sleep Cycle Calculator

Find the best bedtime for your wake-up alarm, or the best wake times if you sleep now. Aligns with 90-minute REM cycles, accounts for sleep-onset latency, sources cited — runs in your browser, no signup.

Open tool

Rate this tool

Be the first to rate

Comments & feedback

Spotted a bug or want an improvement? Tell us — our team reviews every comment, and good ideas get built. Comments are public and anonymous.

Found a bug, edge case, or want to suggest an improvement?

Email me at [email protected] — most fixes ship within 24 hours.