SLI, SLO & Error Budgets
Three terms people use interchangeably
SLI — Service Level Indicator. The measurement. "Proportion of requests served in under 300ms." A number you can compute from your metrics.
SLO — Service Level Objective. A target for that indicator: "99.9% of requests under 300ms over 30 days." It commonly drives internal reliability policy but is not itself a customer contract.
SLA — Service Level Agreement. A contract with customers, often with service credits or other consequences when it is missed. Teams commonly choose a stricter internal SLO to leave operating margin, but that relationship is policy rather than part of either definition.
A good SLI measures something users actually experience. "CPU under 80%" is not an SLI; no user has ever cared about your CPU. "Checkout completes in under 2 seconds" is, because a user notices immediately when it stops being true.
SLI what you measure 99.95% (actual) SLO what you aim for 99.9% (internal) SLA what you promise 99.5% (contractual) A common policy makes SLO stricter than SLA, leaving operating margin before a contract is missed.
3 components2 connections0:00
Recording…