3 Resilience Patterns That Keep a Backend Service Up When Its Dependencies Aren't

A practical breakdown of three resilience patterns for backend services: timeouts, retries with backoff and jitter, and circuit breakers. Using a real-world cascading failure caused by a slow payment provider, the post explains why unbounded waiting kills services, how each pattern bounds a different type of damage, and the correct order to stack them (circuit breaker → retry → timeout → call). Includes Node.js code examples using AbortController, guidance on when to retry (idempotent ops only, not 4xx errors), full jitter vs. naive jitter, and a decision table for which patterns to apply to different call types. Also covers library options like cockatiel and opossum.