SLO / error budget: [Service name]
> SLOs are a shared vocabulary between engineering and product about how reliable the service needs to be — and what happens when it isn't.
SLO / error budget: [Service name]
SLOs are a shared vocabulary between engineering and product about how reliable the service needs to be — and what happens when it isn’t.
Service overview
| Service | |
| Tier | Critical / Important / Best-effort |
| Owner | |
| Dependencies |
SLIs (Service Level Indicators)
| SLI | Definition | Measurement | Source |
|---|---|---|---|
| Availability | % of successful requests / total requests | ||
| Latency (p50) | |||
| Latency (p95) | |||
| Latency (p99) | |||
| Throughput | |||
| Correctness | |||
| Freshness |
SLO targets
| SLI | Target (rolling window) | Window | Burn rate alert |
|---|---|---|---|
| 30d | |||
| 30d |
Error budget
| Monthly budget | 100% - SLO target = available error budget |
| Budget consumed | Tracked in [dashboard / tool] |
| Budget consumed YTD |
When the budget is burning
| Burn rate | Action |
|---|---|
| < 50% / month | Normal ops |
| 50–75% / month | Review; defer non-critical changes |
| > 75% / month | Freeze feature work until budget recovers or SLO is renegotiated |
Exclusions
What does NOT count toward the SLO:
- Pre-production environments
- Planned maintenance with [N]h notice
- Client-side / network errors outside our control
Related
- Monitoring dashboard:
- Alert rules:
- Runbooks:
- Incident retrospective: