Why architecture matters here
SLOs without burn rate alerts + enforcement become paperwork. Architecture matters because the alerts + gates convert measurement into behavior change.
The architecture: every piece explained
The top strip is the primitives. SLI — good events / total. SLO target — 99.9% or whatever. Error budget = 1 - SLO. Burn rate — how fast budget is consumed.
The middle row is alerts. Fast burn alert — 1h × 14.4x consumes 2% budget in an hour. Slow burn — 6h × 6x. Windowed measurement — rolling 28d. Reset policy — monthly.
The lower rows are governance. Enforcement — freeze deploys when budget spent. Postmortem when breached. Ops — SLO council + user metrics + review.
End-to-end flow
End-to-end: SLI = 99.95% success. SLO = 99.9%. Budget = 0.1%. Incident consumes 1.5% budget quickly; fast burn fires. Enforcement freezes risky deploys. Postmortem tracks fix. Budget resets monthly.