What is Uptime?
Also called availability.
Uptime is the proportion of a period during which a system is available for use, usually stated as a percentage. It is commonly expressed in nines, where ninety nine point nine percent allows roughly forty three minutes of downtime per month. The number is only meaningful alongside its definition, since what counts as available and where it is measured vary widely.
The arithmetic is simple and worth memorizing. Ninety nine percent permits about seven hours of downtime per month, ninety nine point nine about forty three minutes, and ninety nine point nine nine about four minutes. Each additional nine reduces the allowance roughly tenfold and typically costs far more to achieve, since it requires removing progressively rarer single points of failure.
The definition matters more than the figure. Availability measured by whether a health endpoint responds is a very different claim from availability measured by whether real user requests succeed. Many published numbers exclude scheduled maintenance, count only complete outages while ignoring severe slowness, or measure at a load balancer that is still healthy when everything behind it is failing.
Partial availability complicates any single number. Modern systems rarely fail completely; more often one feature, one region, or one tenant is affected while the rest works. Reporting a single global percentage can obscure that a small group of customers experienced a total outage. Per feature and per tenant measurement gives a more honest account of what people actually experienced.
Uptime is also a poor sole proxy for user experience. A service that is technically responding but returning stale data, timing out on the operations users care about, or producing incorrect results can score very well on availability while being effectively unusable. This is why availability is usually paired with latency and correctness indicators rather than reported alone.
Key points
- Ninety nine point nine percent is about forty three minutes per month.
- Each additional nine costs far more to achieve.
- The measurement definition matters more than the number.
- A single global figure hides partial and per tenant outages.
- Pair availability with latency and correctness indicators.
In practice
A status page reports ninety nine point nine seven percent uptime for a month, based on a health endpoint probed every minute from three regions. During the same month, one workspace could not run any agent tasks for six hours because of a broken integration specific to its configuration. The global figure stayed high because the probe never touched that path, while for that customer availability was effectively zero.