Uptime is the percentage of time a website or service is reachable and serving correct responses. It is the headline reliability metric and is usually expressed as a number of nines, e.g. 99.9% uptime allows roughly 43 minutes of downtime per month.
Also known as: Availability, Service availability
Uptime is calculated by dividing the time a service successfully responded to monitoring probes by the total time in the measurement window, expressed as a percentage.
A monitoring tool sends a request from one or more locations on a fixed interval (typically every 1 minute for production sites) and records whether the response was a 2xx status code returned within an acceptable time. Sustained failures across multiple regions and retry attempts are then aggregated into incidents, and uptime is the inverse of total incident time.
Different providers count different things as 'down': pure connection failure, slow response (over a threshold), wrong status code, missing keyword in body, or failed TLS handshake. A meaningful uptime number must specify what is being measured.
Uptime is what most SLAs (service level agreements) are written against, and it is the metric customers care about most because it directly maps to revenue lost during outages. Tracking it accurately requires multi-region probes, retry-and-confirm logic to suppress false positives, and clear incident definitions.
See it in the product: Uptime monitoring.