Embrace chaos to achieve stability
summary
A practical case for chaos engineering: a live story about provisioning for a 10,000-person event that creates 4,000 teams, leaves 2–3k Azure DevOps projects in limbo, scales to 150 build agents and triggers a half-day Australia outage. Demos run against a Kubernetes e‑shop: deleting pods causes 404s, scaling to two replicas removes 404s but reveals cart lost state (fixed by Redis state store), and injecting 200ms±100ms latency via Chaos Mesh/Azure Chaos Studio raises request duration and drops order value; automate, minimize blast radius, and run blameless reviews.
AI-written summary · about
chapters (12)
- 0:00Introduction to chaos
- 2:27Chaos theory fundamentals
- 5:03Lessons from scaling
- 9:31Why testing is not enough
- 12:52Experimentation vs testing
- 20:00Principles of chaos engineering
- 23:25Risk management and SLAs
- 28:55Practical experiment design
- 35:00The chaos process
- 37:53Tools and implementation
- 43:23Learning and culture
- 55:04Wrap up and conclusion
related talks
32mImplementing Chaos Engineering in Kubernetes
46mEmbracing Collaborative Chaos: Chaos Days on Large Platforms
33mChaos Engr at Microsoft with Azure Chaos Studio
27mExecuting Chaos Engineering in Production at a Critical Financial Institution
26mBreaking Systems for Resilience
39mChaos: Breaking your Systems to Make them Unbreakable
18mChaos Engineering: Break It On Purpose
40mJourney into the World of Chaos Engineering