Meaning
A Service Level Indicator (SLI) is a quantitative measure of a service's reliability, such as latency percentile or error rate. It addresses the pain point of not having objective data to assess whether a service meets its reliability commitments. Teams reach for an SLI when they need to monitor real‑time performance against agreed targets.
Primary Function
Reliability monitoring
Communicative Purpose
Enables teams to quantify service reliability against agreed targets and trigger corrective actions when thresholds are breached.
Pattern
Define SLO → derive SLI measurement → alert on breach
Core Structure
SLI = good_requests / total_requests
Função primária
Reliability monitoring
Propósito comunicativo
Enables teams to quantify service reliability against agreed targets and trigger corrective actions when thresholds are breached.
Situações de gatilho
Web services: measuring request latency percentiles for API endpoints Microservices: tracking error rate during a new deployment Cloud infrastructure: monitoring uptime of a load balancer
Contextos
Site reliability engineering (SRE), cloud‑native applications, monitoring stacks such as Prometheus, Grafana, and service‑mesh observability frameworks.
Padrão
Define SLO → derive SLI measurement → alert on breach
Estrutura central
SLI = good_requests / total_requests
Colocados típicos
- error budget
- SLO
- latency
- availability
- uptime
Substituições comuns
- Using error rate as an SLI instead of latency percentile (captures failures but hides performance details)
- or using request count ratio (simpler but less granular).
Erros comuns
Treating any request as "good" without filtering out retries → inflates SLI and hides real failures. Calculating latency on the full request stream instead of the tail latency (e.g., 99th percentile) → underestimates latency spikes. Updating the SLI metric less frequently than the service's SLA window → delays detection of degradations.
Similar / contraste
Service Level Objective (SLO): the target value an SLI should meet. Error Budget: the allowable amount of SLI deviation before corrective action. Service Level Agreement (SLA): a formal contract with external customers, often derived from SLOs.
Interferências
Coming from traditional uptime monitoring: assuming 100% availability → ignore latency spikes that can still degrade user experience.
Família do chunk
- service level objective
- error budget
- availability metric
- latency percentile
- request success ratio
Nuance
Do not use an SLI when the underlying metric cannot be reliably measured (e.g., opaque third‑party services). Computing high‑resolution latency percentiles can add CPU overhead; consider sampling for very high‑traffic services. SLI calculations assume a stable request pattern; sudden traffic bursts can temporarily skew ratios.
Efeito pragmático
Correctly defined SLIs give operators early warning of reliability regressions, allowing proactive remediation before customers notice issues.
Dica de memória
Think of an SLI as a service's health‑meter, constantly showing how well it’s performing against its promised speed and reliability.
Nota
An SLI must be observable: the data source should be instrumented and emitted with minimal latency to keep the indicator fresh.
Upgrade path
After mastering SLIs, move to defining Service Level Objectives (SLOs) and managing error budgets.
Log in to save chunks.