Uptime is the share of time a website or service is working and available to the people who use it. It is almost always written as a percentage, and almost always quoted with a lot of nines: 99%, 99.9%, 99.99%. Those numbers sound close together. In practice they are very different, and the gap between them is the gap between a few hours of downtime a month and a few minutes.
This guide explains how uptime is calculated, what each level of “nines” really allows, and how to measure your own.
What uptime means
If your website was reachable for 29 days and 23 hours of a 30-day month, and unavailable for one hour, its uptime for the month is the time it was up divided by the total time:
Uptime % = (total time − downtime) ÷ total time × 100
One hour down in a 30-day month (720 hours) is 719 ÷ 720 = 99.86% uptime.
The opposite of uptime is downtime: the time the service was not working. The two always add up to 100%.
What the nines really allow
This table shows the maximum downtime each uptime level permits. Months are 30 days and years are 365 days.
| Uptime | Downtime per day | Per 30-day month | Per year |
|---|---|---|---|
| 99% | 14 min 24 s | 7 h 12 min | 3 days 15 h 36 min |
| 99.5% | 7 min 12 s | 3 h 36 min | 1 day 19 h 48 min |
| 99.9% | 1 min 26 s | 43 min 12 s | 8 h 45 min 36 s |
| 99.95% | 43 s | 21 min 36 s | 4 h 22 min 48 s |
| 99.99% | 8.6 s | 4 min 19 s | 52 min 34 s |
| 99.999% | 0.9 s | 26 s | 5 min 15 s |
Each extra nine cuts the allowed downtime by a factor of ten, and the cost and effort of achieving it rises with it. Going from 99% to 99.9% might mean choosing a better host. Going from 99.9% to 99.99% usually means redundant servers, automatic failover and a team on call. For most small businesses, shops and agencies, 99.9% is a sensible target.
What counts as downtime
This is where uptime figures quietly differ from one provider to the next. A serious definition counts the site as down when a visitor could not use it, and that includes more than a dead server:
- Error responses, such as a 500 or 503, instead of the page
- Timeouts, where the server takes too long to answer
- Refused or dropped connections
- An expired or invalid SSL certificate, which browsers block with a warning
- A domain that no longer resolves because of a DNS problem
A hosting company’s own uptime promise often excludes scheduled maintenance, problems in your code and anything outside its network. That is why the number on a hosting plan and the number your visitors experience can differ.
How uptime is measured
Uptime cannot be read from your own server, because a server that has failed cannot report that it has. It has to be measured from outside. A monitor requests your site at regular intervals, and each successful response counts as “up”. The percentage is the share of checks that succeeded.
That makes the check interval matter. A monitor that checks every five minutes cannot see an outage shorter than that, and can only estimate the length of one it does catch to within five minutes. For the targets most websites need, a five-minute interval is a good balance. To keep one dropped connection from counting as downtime, Pingvella retries a failed request once straight away before treating the site as down.
Uptime is not the whole story
A site that is technically “up” can still be failing its visitors. A page that takes 15 seconds to load, a checkout that errors while the homepage works, a login that hangs: none of these shows as downtime on a simple uptime check. That is why uptime sits alongside other numbers, such as response time and incident length, in the website monitoring metrics that matter.
It is also why a single percentage hides how the downtime happened. Ten one-minute blips and a single ten-minute outage give the same uptime figure but feel very different to the people affected. The incident history behind the number is just as useful.
Setting a realistic target
- Start by measuring. Run a monitor for a month before promising anything. Your real figure is the starting point.
- Match the target to the stakes. A brochure site can tolerate 99.5%. A shop that takes orders around the clock needs better. A system other people’s software depends on needs more again.
- Count your own changes. A deployment gone wrong is downtime too, and for small teams it is often the biggest source of it.
- Improve the response, not just the host. Hearing about an outage in five minutes instead of two hours does more for your real uptime than almost any infrastructure change.
Measure yours
Pingvella records every website check, logs each incident with its start and end, and shows your uptime over time, so you can see your actual number and where the downtime came from. Every account can have five website checks, including the free one. Read how Pingvella keeps your website alive for how it works, or go straight to website monitoring.
Common questions
What does 99.9% uptime mean?
99.9% uptime allows about 43 minutes of downtime in a 30-day month, or roughly 8 hours 46 minutes in a year. 99% allows about 7 hours a month, and 99.99% about 4 minutes.
How often is my website checked?
Every 5 minutes. If a request fails it is tried once more straight away, so a single dropped connection doesn’t raise a false alarm.
What counts as my website being down?
Anything other than a successful 2xx response once redirects are followed: an error page, a timeout, a refused or dropped connection, an expired or invalid SSL certificate, or an address that no longer resolves.
There are more answers on the FAQ page.