Uptime
Uptime
Uptime is the share of time a site answered when asked. It is the number most people mean when they say a service is up.
How it is measured
Count successful probes over all probes in a window: 43,150 passes out of 43,200 one-minute checks is 99.884 percent. State the probe path, interval, regions, timeout, and window along with the number.
A site can be up in the monitor's eyes and hurting users. If the probe hits a cached page, the number can look perfect while checkout fails.
Worked example
A one-person WordPress hosting side business is monitored by one HTTP check every 5 minutes. In a 30-day month there are 8,640 checks, and 11 fail, so 8,629 pass: 99.873 percent. The monitor implies a 55-minute outage.
Server logs show a 38-minute database crash plus a 6-minute restart, 44 minutes in all. Five-minute spacing overcounts by up to 5 minutes at each edge, so the monitor's 55 is an estimate, not a measurement.
How it differs
Uptime is the good side of the ledger. Downtime is the bad side, in minutes. Uptime excludes how long each failure ran or how many there were. Downtime excludes the denominator that makes it a percentage.
Common errors
Probing a cached URL. Using a single region. Using five-minute intervals for a one-minute promise. Treating 200 as health. Not stating the window. Counting from the host's panel instead of your own probe.
In practice
Add an independent uptime check on a page that touches the database, from two regions, at 1-minute intervals. Record the percentage each month with the definition beside it. Compare it with what your host claims.