metricEngineering & architecture

Service-Level Objective

A service-level objective sets a target value for a service-level indicator over a defined period, expressing the intended reliability level.

Download Markdown (.md)

01What it means

A service-level objective, or SLO, is an internal reliability target attached to a specific service-level indicator and time window. A statement such as “99.9% of eligible requests succeed within 30 days” combines the measured behavior, target, and evaluation period. The permitted shortfall becomes an error budget.[1][2]

Reliability target with a window Attach a target and evaluation period to one trustworthy SLI
SLO = SLI + target + time window
SLI
Measured user-relevant behavior
Target
Required level of good service
Window
Period for evaluation
Error budget
Permitted shortfall from the target

The objective's permitted shortfall becomes the error budget and should trigger predefined action when consumption is unusual.

02Why it matters

An SLO makes “reliable enough” discussable. It helps product, engineering, and operations choose between feature velocity and reliability work, calibrate alerting, and recognize harmful trends before every failure becomes a crisis. A target tied to user needs is more useful than pursuing maximum availability by habit.

03When to use it

Use SLOs for services and journeys whose reliability affects users, revenue, safety, or dependent teams. Establish measurement first, then choose a target informed by user expectations, business consequences, historical performance, and dependency limits. Start simple and revise deliberately.

04How it works

  1. Select a user-centered SLI with trustworthy telemetry.
  2. Choose a target and window that reflect meaningful expectations.
  3. Confirm the objective is achievable given dependencies and architecture.
  4. Track compliance and error-budget consumption, including burn rate.
  5. Define actions for unusual consumption and review the target as needs change.

05A practical example

Example

A document editor sets an objective that 99.95% of save attempts from active sessions complete successfully within one second over 28 days. Alerts focus on rapid budget burn. If failures consume half the monthly allowance in two days, the team pauses a risky storage migration and prioritizes recovery.

06Where it breaks

A single target can hide regional, device, or customer differences. An easy objective becomes decorative, while an impossible one creates permanent emergency. SLOs do not replace capacity planning, security, qualitative research, or judgment, and they should not be confused automatically with customer contracts.

Common mistakes

  • Choosing a target before defining and validating the indicator.
  • Setting 100% as the default without considering cost or feasibility.
  • Reporting compliance without an agreed response to budget consumption.

07Sources

  1. Service Level ObjectivesGoogle Site Reliability Engineering · Foundational book chapterService Level Objectives
  2. Implementing SLOsGoogle Site Reliability Engineering · Institutional implementation guideImplementing SLOs
View saved topics

Start typing to search titles, definitions, categories, and frameworks.