Resilience Engineering
summary
Maps direct and indirect dependencies, then applies isolation, redundancy and graceful degradation to limit failure blast radius. Uses destructive testing and fault injection, plus end-to-end tests instrumented across about 10 systems to reproduce real conditions. For Wallaroo the team runs topology generators, fuzzes system events (node add/remove, crashes, network faults) and validates per-key no loss, no duplication, no reordering, and reports found ~200 edge-case bugs from ~100 tests before production.
AI-written summary · about
chapters (14)
- 0:00Start
- 2:11Failure is Going to Happen
- 3:20Tools & Methods
- 4:34Mapping Dependencies
- 6:06Remove Unnecessary Coupling
- 6:48Hedging
- 8:17Graceful Degradation
- 9:26Destructive Testing
- 10:46End-to-end Testing
- 15:11Property-based testing
- 16:32Fuzzer
- 18:50Fuzzing Foobr
- 19:45Fuzzing in Wallaroo Tests
- 22:48Key Approaches
related talks
40mResilience Testing - Failure is an Option
05mWhat is resilience engineering
16mEngineering Failure-Resilient Systems
22mResilience Is Not Enough
32mWhat is resilience engineering?
51mWithout Resilience, Nothing Else Matters
10mDesigning for Failure using AWS: Resilient Systems
44mFrom here to resilience - a travel guide