FastAPI Performance and Resilience

When to use

Invoke when a scaffolded FastAPI service must hold an SLO under load: blocking calls are stalling the event loop, downstream failures cascade, the connection pool is mis-sized, there is no caching posture or backpressure under burst, retries amplify outages, or there is no load-test gate proving the service meets its latency/throughput budget before release.

Do not use for: the service shell, settings, or error tiers (use fastapi-service-scaffold), auth or OWASP review (use fastapi-auth-and-security-review), OpenTelemetry/metrics/SLO definition (use fastapi-observability-readiness), or broker/task/event integration (use fastapi-async-and-task-integration).

Inputs

Required:

• A service with the fastapi-service-scaffold baseline (settings, logging, ASGI-lifespan shutdown, health probes present).
• Approved architecture/performance budgets and architecture/reliability SLO targets (latency/throughput objective, error budget) — or explicit confirmation they are intentionally deferred.

Optional: