Let it fail: resilience patterns for microservices
summary
FREE NOW runs ~250 services and shows concrete resilience wins: a voucher outage costs ~100k€ and once causes a missed flight, so the platform adopts retries with capped attempts, exponential backoff and randomized jitter, Spring Retry for JVM, Netflix Hystrix (moving to Resilience4j), and per-operation metrics for root-cause. Engineering implements circuit breakers, bulkheads (thread-pool isolation) and load‑sharing, idempotence checks to avoid double charges, health endpoints, and tracing; result: fewer fires, less alert fatigue, and recoveries without redeploys.
AI-written summary · about
chapters (10)
- 0:00Introduction and real story
- 2:15What is resilience in microservices
- 5:53Retry patterns and pitfalls
- 9:28Exponential backoff with jitter
- 14:55Idempotence in distributed systems
- 17:30Graceful degradation with circuit breakers
- 24:30Bulkheading for isolation
- 29:53Load shedding in practice
- 30:18Health checks and endpoints
- 32:07Summary and benefits of resilience
related talks
49mManaging Resiliency in Microservices
50mSurviving Scale: Patterns for Evolving Failure-Resilient Microservices
44mArchitectural Patterns of Resilient Distributed Systems
41mResillient Microservices without the Chaos
51mDesigning for Failure: Build Resilient Services
25mResilience testing in modernization projects
32mBeyond Optimization: Resilient Cloud Microservices at Scale
35mSCALA BASED SERVICES RESILIENCE IN THE REAL WORLD