Free tool Uptime calculator: SLA % to allowed downtime
What does 99.9% actually allow? Enter an availability and see the downtime budget for each period, enter an outage and see what it did to your monthly figure, or list your components to see what the whole chain can promise.
- Both directions
- Series and parallel
- Runs locally
Convert uptime and downtime
Pick a direction. A month is 1/12 of a year (30.42 days) unless you choose a fixed length — check which one your SLA uses.
Used for the per-month figures.
e.g. 99.9 or 99.95%.
e.g. 43m, 2h 30m, 1:30:00. Plain numbers are minutes.
One per line, in order: name availability. Add x2 for two redundant copies in parallel. All lines are in series.
Availability
99.9%
Number of nines
3.0
−log₁₀(1 − availability).
Components
Stage availability = 1 − (1 − availability)^copies. The weakest stage is highlighted.
Allowed downtime
The downtime budget at this availability.
| Period | Allowed downtime |
|---|---|
| Per day | 1 min 26 s |
| Per week | 10 min 05 s |
| Per month | 43 min 48 s |
| Per quarter | 2 h 11 min |
| Per year | 8 h 46 min |
Nines table
Allowed downtime for common SLA levels.
| Availability | Per day | Per week | Per month | Per quarter | Per year |
|---|---|---|---|---|---|
| 90% | 2 h 24 min | 16 h 48 min | 3 d 01 h | 9 d 03 h | 36 d 12 h |
| 95% | 1 h 12 min | 8 h 24 min | 1 d 13 h | 4 d 14 h | 18 d 06 h |
| 99% | 14 min 24 s | 1 h 41 min | 7 h 18 min | 21 h 54 min | 3 d 16 h |
| 99.5% | 7 min 12 s | 50 min 24 s | 3 h 39 min | 10 h 57 min | 1 d 20 h |
| 99.9% | 1 min 26 s | 10 min 05 s | 43 min 48 s | 2 h 11 min | 8 h 46 min |
| 99.95% | 43.2 s | 5 min 02 s | 21 min 54 s | 1 h 06 min | 4 h 23 min |
| 99.99% | 8.64 s | 1 min | 4 min 23 s | 13 min 08 s | 52 min 34 s |
| 99.999% | 864 ms | 6.05 s | 26.3 s | 1 min 19 s | 5 min 15 s |
| 99.9999% | 86 ms | 605 ms | 2.63 s | 7.88 s | 31.5 s |
Share these inputs: the values go after the # in the link, so they never reach a server.
https://tapcub.com/tools/uptime-calculator
Version v1.0Updated Update history
What an availability number commits you to
Each extra nine cuts the downtime budget by ten times. The table makes that concrete.
Allowed downtime
At 99.9%, the budget is 43 min 48 s a month or 8 h 46 min a year. One unplanned deploy rollback can use a month's budget.
Number of nines
99.9% is three nines, 99.95% is about 3.3. Use it to compare levels quickly; it says nothing about how outages are spread.
Downtime → availability
A 2.5-hour outage in one quarter leaves 99.89% for that quarter. Whether it breaches the SLA depends on the period the SLA is measured over.
Composite availability
Components in series multiply, so the chain is always weaker than its weakest part. Redundant copies in parallel push a stage much higher — if they really fail independently.
Availability says nothing about when downtime happens. Thirty seconds a day and one 3.5-hour outage a year are both about 99.96%, and customers feel them very differently.
Three questions it answers
The same arithmetic, read in three directions.
-
1
Writing or reading an SLA
Enter the promised percentage to see the downtime budget per month and year, with the month length the contract uses.
-
2
After an incident
Enter how long it lasted and the measurement period to see the availability you actually delivered.
-
3
Designing a system
List components in order and mark redundant ones with x2 or x3 to see what the whole stack can realistically promise.
The arithmetic behind the nines
Availability is uptime ÷ total time. Everything else follows from that.
- Year = 365 days; month = year ÷ 12 (30.42 days) unless set; quarter = year ÷ 4; week = 7 days
- Series: every component must work, so availabilities multiply
- Parallel: the stage fails only if every copy fails at once, assuming independent failures
- Planned maintenance counts as downtime unless the SLA excludes it
Allowed downtime
downtime = (1 − availability) × period
(1 − 0.999) × 30.42 days = 43 min 48 s
Availability from downtime
availability = 1 − downtime ÷ period
1 − 2.5 h ÷ 2,190 h = 99.886%
Series
A = A₁ × A₂ × … × Aₙ
99.9% × 99.95% = 99.85%
Parallel (n copies)
A = 1 − (1 − A₁)ⁿ
1 − (1 − 0.999)² = 99.9999%
What redundancy buys
An app server at 99.9% and a database at 99.95%, first single, then with a second copy of each. Assumes the copies fail independently.
| Setup | Availability | Downtime per year |
|---|---|---|
| App server only | 99.9% | 8 h 46 min |
| App server + database, single | 99.85% | 13 h 08 min |
| Both doubled (x2) | 99.999875% | 39.4 s |
Adding a database in series makes things worse; doubling both makes them dramatically better — on paper. Shared power, network, deploys and configuration make real failures far less independent.
Where uptime numbers mislead
The arithmetic is exact. The definitions around it are not.
Different month lengths
A 28-day February allows less downtime than a 31-day month at the same percentage. Use the length your contract uses.
Forgetting dependencies
Your service is only as available as everything in series with it: DNS, cloud region, payment provider.
Assuming independent copies
Two servers behind the same switch or config fail together. The parallel formula is a best case.
Leaving out maintenance
Planned downtime still counts for users unless the SLA explicitly excludes it.
TapCub does not monitor uptime
This calculator is a general conversion tool; TapCub has no uptime checks or status pages. What its website analytics can show is how visitors behave around an incident.
Realtime visitors
A live list of visitors on the site. A sudden drop is a symptom worth checking, not a substitute for an uptime check.
Learn moreAlerts on metrics
Rules for a drop against the previous period, a threshold or an anomaly, sent by email, WeCom, DingTalk, Feishu or webhook.
Learn moreSpeed and errors
Real-user Core Web Vitals and JavaScript, resource and API errors with the sessions they affected.
Learn moreUptime questions
Still have a question?
Chat with the team behind TapCub — we usually reply within a few hours on business days.
How much downtime is 99.9% uptime?
About 1 min 26 s a day, 10 min 05 s a week, 43 min 48 s in an average month and 8 h 46 min a year. With a 30-day month it is 43 min 12 s.
How do I calculate uptime percentage?
Divide the time the service was up by the total time in the period and multiply by 100. A 30-minute outage in a 30-day month is 1 − 30 ÷ 43,200 = 99.93%.
What does "five nines" mean?
99.999% availability: about 5 min 15 s of downtime a year, or 26 seconds a month. Each extra nine divides the downtime budget by ten.
How do I calculate availability of components in series?
Multiply their availabilities. Two components at 99.9% each give 99.8%. The result is always lower than the weakest component.
How does redundancy change availability?
For n independent copies, availability is 1 − (1 − A)ⁿ. Two copies at 99% give 99.99%. Real copies share failure causes, so treat this as an upper bound.
Does TapCub monitor my website uptime?
No. TapCub is website analytics and live chat. Its realtime view and metric alerts can show a sudden drop in visitors, but they are not uptime checks; use a dedicated monitoring service for that.
Which month length should I use?
Whatever your SLA defines. Many measure per calendar month; others use an average month or a fixed 30 days. When customers are in several time zones, the time zone converter helps agree on where a day starts.
Update history
New features, fixes and data refreshes for this tool, newest first.
-
· v1.0
LaunchedFirst release.
Missing something or found an error? Tell us — requests that come up often go into the next update. All tool updates
See how visitors behave around an incident
Uptime tells you whether the site responded. Analytics tells you what visitors did: how many left, which pages failed, which errors they hit.