Performing Chaos in a Serverless World
summary
Serverless chaos experiments use concrete tools and numbers: wrapping Node.js handlers with failure-lambda (or Python chaos-lambda), storing JSON configs in Parameter Store, injecting latency between 100–400 ms, returning status codes like 404 or 502, or applying a denylist to block DynamoDB/S3. A circuit breaker opens after a threshold of 3 failures and serves a static fallback image. Experiments run as a canary 10% step in CI/CD so failures affect a small blast radius and trigger rollback if metrics exceed thresholds.
AI-written summary · about
chapters (13)
- 0:00Start
- 0:44Abstract
- 1:07Agenda
- 1:40About me
- 2:23What is chaos engineering?
- 4:16Motivations behind chaos engineering
- 7:42Serverless chaos experiments
- 14:31Failure-lambda NodeJS
- 16:54Serverless chaos demo
- 34:48What's next?
- 35:33Serverless chaos CI/CD demo
- 47:09Summary
- 47:54Do you want more?
related talks
24mBreaking Serverless Things on Purpose: Chaos Engineering in Stateless Environments
34mChaos engineering for serverless
44mApplying principles of chaos engineering to serverless
43mApplying the Principles of Chaos to Serverless
19mBuilding Resilient Systems - Serverless WebDev
25mHow to Fail at Serverless (without even trying)
51mConfusion In The Land Of The Serverless
39mBuilding Resilient Serverless Applications