A service-level objective (SLO) defines a target for a measurable service outcome over a time window. It creates a shared reliability goal for engineering and product teams.
Choose a user-centered indicator
Examples include successful requests, timely completion of a background workflow, or availability of a critical page. Define the event population precisely so exclusions cannot be changed casually during an incident.
Set a realistic target
A 99.9% objective is not automatically appropriate. Use historical performance, business impact, dependency limits, and user expectations. A target that is too strict can consume resources without meaningful benefit.
Use error budgets
The error budget is the permitted unreliability within the SLO window. Teams can use burn rates to detect when failures are consuming the budget too quickly. A depleted budget may trigger a pause on risky releases until reliability improves.
Review the objective
SLOs should evolve as service importance and architecture change. Keep the objective, measurement query, and response policy documented together.
Checklist
- Define a measurable user-facing indicator.
- Specify the population and time window.
- Set a justified target.
- Monitor error-budget burn rate.
- Agree on actions when the budget is exhausted.