What is Uptime?

Uptime is a metric that expresses the proportion of time a service, server, or network component remains operational and reachable. It is almost always stated as a percentage, with common targets being 99.9%, 99.99%, or 99.999%. The calculation excludes planned maintenance windows if a service level agreement defines them, but otherwise counts any period of unavailability. A service with 99.9% uptime over one year can be unreachable for about 8 hours 45 minutes. At 99.99% the allowed downtime drops to roughly 52 minutes per year. These numbers are direct consequences of basic arithmetic: (1 - X) multiplied by 8760 hours in a typical year.

How uptime is measured depends on the monitoring setup. Providers often check from multiple geographic locations using ICMP pings, TCP port checks, or HTTP request probes. A single failed check may not count as downtime; most agreements require consecutive failures over a defined window before declaring an outage. The measurement period is usually calendar months, though annual uptime is also quoted. Credits or refunds are commonly tied to falling below a guaranteed threshold, for example 99.9%.

Uptime is not a measure of performance. A server can be "up" with high ping times or low throughput. It also says nothing about data integrity or security. In the hosting stack, uptime is a basic contractual floor. Infrastructure engineers treat it as a lagging indicator; they focus instead on mean time between failures (MTBF) and redundancy through load balancers, failover clusters, and multi-AZ deployments. Uptime alone does not capture partial failures, degraded states, or silent errors, which is why modern observability frameworks favor service-level objectives (SLOs) based on latency and error rate.