The fraction of time a service is in a working state, the simplest reliability number a customer ever asks about.
Uptime is the percentage of time a service is up and serving correctly during a window. It's expressed in 'nines': 99% (3.65 days down/year), 99.9% (8.76 hours), 99.99% (52 minutes), 99.999% (5 minutes). Uptime is what status pages report, what SLAs commit to, and what customers actually feel. The internal counterpart, SLO availability, is usually one nine stricter than the public SLA so the team has buffer for routine misses.
Uptime is the most plain-language reliability number, and it's what executives, customers, and procurement teams understand without translation. Engineering tracks SLOs and error budgets internally, but every external conversation converts to uptime. Knowing the cost ladder between nines (each nine is roughly 10x the engineering investment) is the key to setting a target that matches the service tier.
See the part of the platform that handles uptime in production.