Design Patterns for Building Resilient Systems - CodeOpinion

Abstraction: Fallbacks, retries, and circuit breakers applied per-context for resilient systems

Key points:

• CrowdStrike 2024 incident prompted re-evaluation: author detected indirect impact via fallback metrics firing on third-party API failures
• Three core patterns — fallbacks, retries with backoff, and circuit breakers — should be composed together, never applied in isolation (retry alone adds dangerous latency)
• Fallback values should be "good enough": e.g. showing USD price when exchange-rate API is down rather than breaking the page
• Context is critical: sync client-facing requests require tight timeouts and low-latency fallbacks; async message-queue workers have very different latency tolerances and retry windows
• Circuit breaker pattern: after N failures in a time window, skip the primary service and go directly to fallback until the service recovers
• Bulkhead pattern separates degrading routes from healthy ones to prevent cascading failures across the whole system

Connections: Resilience Patterns · Distributed Systems · Fault Tolerance · Crowdstrike

Source: https://codeopinion.com/design-patterns-for-building-resilient-systems/