SLI/SLO Definition Framework
🔧 Purpose
This framework helps product, engineering, and SRE teams clearly define, measure, and track service quality. It turns service health into actionable metrics that drive both product decisions and operational monitoring.
📝 Core Concepts
Term
Meaning
SLI (Service Level Indicator)
What you measure — a specific performance or reliability metric.
SLO (Service Level Objective)
The target goal for that metric.
SLA (Service Level Agreement)
The external customer-facing promise based on internal SLOs.
📄 SLI / SLO Definition Template
1️⃣ Service / Function Name
e.g. Document Load, File Upload, Search API
2️⃣ SLI — Service Level Indicator
Indicator Name:
e.g. Document Load Time
Definition:
What is measured?
Exact measurement start & end points.
e.g. Time from first request received to first contentful paint.
Measurement Method:
e.g. Collected via synthetic monitoring and backend request logs.
3️⃣ SLO — Service Level Objective
Target Value:
e.g. 99% of document loads complete within 1 second
Measurement Window:
e.g. Rolling 30 days
Error Budget:
e.g. 1% allowed outside target per window
4️⃣ Dependencies
What systems or services does this depend on?
e.g. CDN, API Gateway, Database
5️⃣ Business Impact
What happens if this SLO is not met?
e.g. Poor user experience, drop in active users, revenue loss
6️⃣ Monitoring & Alerting
Monitoring Tool:
e.g. Datadog, Grafana, Prometheus
Alert Threshold:
e.g. Alert if 95th percentile latency exceeds 1 second for more than 15 minutes
7️⃣ Ownership
Who owns this metric?
e.g. Platform Engineering Team
8️⃣ Related SLIs
List any connected SLIs for this service
e.g. API Error Rate, Upload Latency, Query Timeout
9️⃣ Open Questions / Notes
Capture any pending assumptions, known data gaps or future revisions.
🔑 When to define SLI/SLO:
New feature launches
Major infra or architecture changes
Critical customer-facing flows
Core platform services
💡 Pro tip for teams:
Start with 3–5 high-impact SLIs.Don't over-engineer early on — prioritize metrics that reflect real user experience.