CONCEPT Cited by 1 source
Service Level Indicator (SLI)¶
A Service Level Indicator (SLI) is the measurement a Service Level Objective is defined over — the actual metric you collect. Typical SLIs are request success rate, request latency, queue freshness, throughput, and availability, each computed over a rolling window and compared to the SLO target.
SLI/SLO pairs originated in the Google SRE book (2016) and became the standard reliability-measurement primitive across the industry.
The measurement-definition problem¶
The hard part of defining an SLI is picking the measurement that reflects user-visible behaviour rather than internal component health. A service can be "up" (process alive, health check green) while still failing its SLO (queue lag too high, 99th-percentile latency blown, error rate past the budget). SLIs on user-path metrics catch this; SLIs on liveness don't.
Seen in¶
- sources/2021-09-12-zalando-tracing-sres-journey-in-zalando-part-i — Zalando rolled out SLIs and SLOs together as the baseline reliability-measurement primitive, accompanied by the first SLO Reporting tool (SLR).