Incident Analysis Before Chaos Engineering
summary
Builds and runs a production chaos platform used at Netflix to inject failures, discovers the real value comes from mining past incidents to prioritize experiments and surface organizational gaps. Uses chaos tooling at Netflix plus automated experiment selection, runs third‑party investigations and cognitive interviews to reveal surprising facts: a new on‑call person owned two systems at 3am, promotion packet deadlines cause incident spikes, tribal knowledge rescues incidents. Recommends a facilitated post‑incident pipeline and metrics beyond MTTR.
AI-written summary · about
chapters (10)
- 0:00Intro
- 5:45Cases & stories
- 6:00Story 1: Make chaos engineering tooling work better
- 9:33The secret
- 10:51Story 2: It wasn't just "human error"
- 16:13Story 3: Promotion packets were due
- 18:50What can good incident analysis help you with?
- 20:36When are we ready for "incident analysis" & what to do today?
- 21:50How do you know if it's working?
- 24:00Outro
related talks
38mChaos Engineering: A Step Towards Resilience
38mChaos Engineering
52mChaos engineering: Breaking things on purpose in production
49mChaos & Intuition Engineering at Netflix
36mChaos Engineering
18mChaos Engineering: Break It On Purpose
48mChaos Engineering: The Path to Reliability
26mPsychology of Chaos Engineering