SLOs

A service level objective, or SLO, is a concrete reliability target for a service, such as “99.9% of requests succeed within 300 ms, measured over 30 days”. It’s built on an SLI, the indicator you actually measure, like the fraction of good requests. The useful twist is the error budget: a 99.9% target means 0.1% failure is allowed, about 43 minutes of downtime a month, and that allowance is something to spend. While budget remains, teams ship features and take risks; when it’s burning too fast, they slow releases and work on reliability. SLOs turn “is the site reliable enough?” from an argument into a number both engineers and product owners can agree on.