A service level indicator (SLI) is a quantitative measurement of behavior users care about, such as successful-request ratio, request latency or durable job completion.

An SLO sets a target for an SLI over a defined window. For example:

At least 99.9% of eligible requests succeed over 30 days.

The denominator matters. Counting internal health probes together with user requests can produce a precise number that does not represent user experience.

Latency objectives are commonly expressed as a fraction of requests below a threshold rather than only an average, because a small number of very slow requests can be hidden by the mean.

An SLO is an engineering target, not a claim that failures are impossible. The target should reflect what users require and what the system is designed to deliver.

SLIs turn reliability into measurable behavior; SLOs state how much of that behavior is considered acceptable.