Cloudy with a Chance of Chaos
summary
Vanguard builds a homegrown chaos platform called Climate of Chaos that runs experiments named Cyclone/Tornado/Blizzard and now supports 16 fault types targeting EC2, ECS and RDS with EKS planned. The system uses serverless (Lambda + Step Functions) plus a Flask front end on Fargate, DynamoDB on‑demand, CloudWatch events and Splunk dashboards. Experiments are single‑target, tag‑gated via IAM, run in non‑prod, include a synthetic probe called Radar, and cut costs from hundreds to cents by using consumption pricing.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:57What is Chaos Engineering?
- 10:04Why is Chaos Engineering Valuable?
- 12:15Build vs. Buy
- 14:14The Climate of Chaos
- 15:45Our Guiding Principles
- 16:05Serverless Architecture
- 23:46Climate Control State Machine
- 25:08Easy Adoption
- 27:13Defined Guardrails
- 30:38Integrated Reporting
- 32:47What's next for our team?
- 37:14Thank you!!
related talks
55mChaos Engineering for Cloud Foundry Platform
57mChaos Engineering Primer
01h 08mDystopia as a Service
33mImprove resilience with automated chaos engineering
49mChaos & Intuition Engineering at Netflix
42mChaos Engineering: Why Breaking Things Should be Practiced
18mFostering Chaos Engineering Culture
35mChaos Engineering WG Deep Dive