SLIs

An SLI, or service level indicator, is a carefully defined quantitative measure of a specific aspect of a service's quality. It is used to assess how well a system performs relative to defined expectations, typically from the perspective of users. SLIs are derived from telemetry data collected from real user traffic or synthetic tests and are chosen to be observable, measurable, and meaningful to stakeholders.

SLIs are often used within a broader reliability framework that includes service level objectives (SLOs) and service level agreements (SLAs). An SLO is a target value or range for one or more SLIs, such as “99.9% availability” or “p95 latency under 250 ms.” An SLA is a formal contract with consequences if targets are not met, usually specifying remedies or penalties. While SLIs are the metrics, SLOs are the performance goals, and SLAs formalize commitments to customers or partners.

Common SLIs cover availability, latency, error rate, and saturation. Availability measures the proportion of time the service functions as expected; latency tracks response times; error rate monitors the share of failed requests; saturation reflects resource pressure or throughput limits. Measurements may use real-user data, synthetic tests, or a combination, and are often calculated over defined windows (for example, a 30-day rolling window or a 14-day period).

Practically, teams use SLIs to drive decisions through error budgets, alerts, and dashboards. An error budget expresses the permissible level of unreliability within an SLO and informs release and incident-management priorities. Good SLIs are tightly aligned with customer impact, challenging enough to indicate meaningful reliability, and supported by high-quality instrumentation. Common pitfalls include choosing misleading metrics, overloading with too many SLIs, or letting data quality degrade.