SLI, SLO & SLA

Defining and measuring service reliability through Service Level Indicators, Objectives, and Agreements with error budgets.

Overview

Overview SLI (Service Level Indicator), SLO (Service Level Objective), and SLA (Service Level Agreement) form a hierarchy for defining, measuring, and communicating service reliability. Originating from Google's Site Reliability Engineering (SRE) practice, these concepts provide a data-driven framework for making trade-offs between feature velocity and reliability investment.

An SLI is a quantitative measure of some aspect of service health — such as the proportion of requests that return successfully within 200ms. An SLO is the target value for that SLI — for example, "99.9% of requests must succeed within 200ms, measured over a 30-day rolling window." An SLA is a formal business contract that specifies consequences (usually financial penalties) if the service fails to meet certain SLOs. The relationship is: SLI measures reality, SLO sets the engineering target, and SLA sets the business commitment (typically less aggressive than the SLO to provide a buffer).

The most powerful concept in this framework is the error budget: the difference between 100% and the SLO target. A 99.9% availability SLO means 0.1% of requests can fail — that is the error budget. When the budget is healthy, teams can ship features aggressively (accepting some risk). When the budget is nearly exhausted, teams shift focus to reliability work. This transforms the "reliability vs velocity" debate from a political argument into a data-driven decision.

Key concepts

  • SLI (Service Level Indicator) — what you measure
  • SLO (Service Level Objective) — what you target
  • SLA (Service Level Agreement) — what you promise
  • Error budgets and budget burn rate
  • The four golden signals (latency, traffic, errors, saturation)
  • Percentile-based SLIs (p50, p99, p99.9)
  • SLO-based alerting vs threshold alerting

Interview questions

  • What is the relationship between SLI, SLO, and SLA?
  • How do error budgets work and how do they drive engineering decisions?
  • What are the four golden signals and how do you implement them?
  • How do you set appropriate SLOs for a new service?

Sign up free to read the full topic

More System Design topics

Start free on Cariara · Coding problems · System design studio