Harnessing Chaos; the hidden ingredient behind building better systems
summary
Production is hostile and unpredictable so the talk argues for deliberately experimenting to raise availability rather than avoiding failure. It specifies practices: run limited-scope continuous disaster recovery (game days, start small blast radius), require good observability, automate experiments, and run them in production when safe. It demos the open source chaos toolkit (Python) with a failing “exchange file” example and shows iterative fixes; real wins include shortening recovery time — one system is recreated in 45 minutes and then hardened.
AI-written summary · about
chapters (13)
- 0:00Start
- 8:19Introduction of Myself
- 15:12Chaos Engineering Has Nothing To Do with Chaos
- 20:36Game Days
- 21:22Chaos Engineering
- 23:08How Do You Do Chaos Engineering
- 28:30The Best Way To Start Chaos Engineering
- 28:56The Best Place To Run Chaos Experiments Is in the Real World
- 32:26Chaos Run Command
- 34:46The Closest Testing Technique to Chaos Engineering
- 34:58Steady State Hypothesis
- 38:20Observability
- 39:00Rules of Chaos Club
related talks
01h 04mHarnessing Chaos; the hidden ingredient behind building better systems through learning
26mBreaking Systems for Resilience
01h 04mTrust and Confidence through Chaos
37mChaos Engineering: The Fine Art of Breaking Stuff in Production on Purpose
01h 02mPractical Chaos Engineering breaking things on purpose to make them more resilient against failure
43mBetter Software Development Systems through Stress
18mChaos Engineering: Break It On Purpose
23mHow to tame chaos effectively?