Uptime: definition
Uptime is the period or proportion of time that a system is considered operational under a defined check. For websites, that check often combines reachability, an expected HTTP status and sometimes content or transaction validation.
Uptime is a measurement rule, not just a number
A monitor has to decide what “up” means. A TCP connection can succeed while the application returns an error. An HTTP 200 response can contain an error page. A homepage can work while login or checkout is broken. The stronger the business outcome, the more the check should validate that outcome directly.
How uptime is commonly calculated
For a simple time-based calculation, subtract downtime from the total observation period and divide by the total period. Scheduled checks estimate the state between observations, so check interval and confirmation logic influence the result.
Probe location and confirmation
One probe can confuse a local routing problem with a global outage. Production monitoring often uses multiple locations or a confirmation step before declaring downtime. Regional services may need availability measured separately by geography.
What an uptime monitor should record
- Timestamp and check location.
- DNS, connection and response timing where available.
- HTTP status and any content assertion.
- Timeout or network error details.
- Incident start, confirmation and recovery time.
- The service owner and escalation path.
Uptime vs availability
The terms overlap, but availability is often broader. Uptime can describe whether a target appears operational, while a service availability indicator may count successful user requests or transactions. Define the term used in your SLO or SLA instead of relying on a generic percentage.
Related concepts
Short answer
Uptime measures how often a target is operational according to a specified check. Define what counts as “up,” use appropriate probe locations and confirmation logic, and keep raw incident evidence behind the percentage.
Practical checklist
- Define what “up” means beyond simple reachability.
- Use representative probe locations and confirmation logic.
- Preserve the exact failure, check time and recovery time.
- Separate external detection from internal root-cause analysis.