SLIs, SLOs and SLAs — defining, instrumenting and communicating — SRE, Platform Engineering & Professional Readiness

Before a team can manage reliability, it must answer a deceptively hard question: what does 'the service is working' actually mean in measurable terms? Service Level Indicators (SLIs), Service Level Objectives (SLOs), and Service Level Agreements (SLAs) are the three-layer framework that turns that question into an actionable reliability programme.

An SLI is the raw measurement — request success rate, P99 latency, data freshness age. An SLO is the target range for that measurement. An SLA is the external contract that specifies commercial consequences when the service falls below the committed threshold. Without this framework, reliability is managed by intuition, escalation, and political pressure.

With it, every reliability decision — how much redundancy to build, whether to release a risky change, when to declare an incident — becomes grounded in quantified obligations rather than subjective confidence. Understanding how to define, instrument, and communicate all three layers is foundational to every other SRE discipline.

Analogy🏏Cricket

🏏 Think of it like cricket: The SLI/SLO/SLA framework maps directly onto how a cricket franchise manages player performance. The SLI is the raw measurement from the scorecard: batting average, strike rate, economy rate — observable, precise, agreed-upon numbers. The SLO is the coaching staff's internal target: 'our number-three batsman should maintain a minimum strike rate of 130 in T20 matches.' The SLA is the franchise's contract with the board: 'if the squad averages below 7.0 runs per over across 20 matches, the franchise pays a performance penalty.' Just as the coaching staff monitors the SLO target daily to intervene before the SLA consequence triggers, SRE teams monitor SLOs to act before contractual SLAs are breached. The insight: SLIs, SLOs, and SLAs are not bureaucratic formalities — they are the measurement infrastructure that makes reliability a manageable engineering problem.