SLO

SLO stands for service level objective. In reliability engineering, an SLO is a specific, measurable target for a service's performance over a defined period. An SLO is typically stated in terms of an SLI, or service level indicator, and is used to assess whether the service meets its expected quality.

SLOs are distinct from SLAs. An SLI measures a property of the service, such as availability or latency. The SLO sets a target for that measure. A service level agreement, when present, translates one or more SLOs into contractual obligations and may include remedies or penalties for failure.

Defining and measuring SLOs involves selecting one or more SLIs, choosing a target, and specifying a time window for evaluation. Common targets are percentages or percentiles over a period, and calculations are based on data from monitoring, tracing, and logs. The method should be documented and reproducible.

Common SLO domains include availability, latency, error rate, saturation, and data freshness. Examples: 99.9 percent availability in a 30-day window; p95 latency under 200 milliseconds; error rate below 0.1 percent.

An example SLO clarifies expectations; for instance, an online service may target 99.9 percent uptime per month, allowing at most about 43 minutes of downtime.

Error budgets and burn rates are used to balance reliability and velocity. The error budget equals 1 minus the SLO. When burn rate is high, teams may slow feature deployment or invest more in reliability.

Implementation considerations include automating measurement, building dashboards, and setting alerting that reflects the SLOs. SLOs should be aligned with business goals, reviewed regularly, and understood by stakeholders across product, engineering, and operations.