2025-01-22 · Aisha Rahman

Error Budgets Without the Theater

SLO slides often look precise and act vague. Teams pick 99.9% because it sounds serious, then ignore burn rates until an incident.

Start with one user journey and one SLI you can measure today — usually availability or success rate on a critical API. Set a target you can afford to miss occasionally; error budgets exist to prioritize reliability work, not to punish teams.

Alert on burn rate, not static thresholds on raw metrics. Two-window policies catch fast burns and slow leaks. Document what happens when budget is exhausted: feature freeze, reliability sprint, or explicit risk acceptance.

Observability for System Designers walks through this with worksheets, not vendor pitches.

#SLO #observability