reliability-observability-gate

To define SLI SLO impact, you specify SLIs for user-facing paths and set error budgets that must be respected before release. This ensures production changes have measurable reliability targets and bounded failure modes.

Incident readiness requires structured logs with trace context propagation, bounded metric label cardinality, and multi-window multi-burn-rate alerting. These telemetry requirements ensure alerts are actionable and observable during incidents.

Validating resilience controls requires checking circuit breakers, rate limits, timeouts, retries, fallbacks, and DLQ depth monitoring. Explicit recovery and rollback criteria must also be tested for changes involving external dependencies or async jobs.

Yes, this reliability gate applies to database-impacting and infrastructure-only behavior-risk changes. It captures latency budgets, capacity guardrails, saturation signals, and cost exposure to ensure safe production deployments.

Multi-window multi-burn-rate alerting is a method to detect rapid and slow error budget consumption across different time windows. You need it to ensure observability and timely incident alerts for critical production changes.

To set capacity planning and cost guardrails, you verify latency budgets, concurrency limits, saturation signals, and cost exposure. Capturing these capacity guardrails explicitly prevents saturation and unbounded costs during production deployments.