Chaos Engr. in the Fast Lane
summary
Claim: kernel‑level eBPF instrumentation plus generative AI lets teams run reproducible, security‑aware chaos experiments without changing apps or rebooting. The demo uses Tetragon/Cilium and a Python loader to attach eBPF programs into the kernel, traces system calls and TCP flows for a boutique‑shop microservice (checkout target), injects CPU load and capability/process probes, emits JSON events to CI/CD, and feeds an AI‑powered chaos copilot (OpenAI) to generate hypotheses. Experiments detect and replay real‑world behaviors and succeed at reproducing incidents.
AI-written summary · about
chapters (17)
- 0:00intro
- 0:22preamble
- 0:33about francesco and michele
- 1:11agenda
- 5:10what is ebpf?
- 6:17how does ebpf work?
- 9:18augmenting chaos engineerign with ebpf
- 11:52lessons learned - key challeges & trends
- 13:08augmenting chaos engineering with ebpf
- 14:45design enhanced security chaos experiments within kubernetes
- 15:21potential security k8s vulnerabilities
- 17:27advanced chaos experiments with ebpf
- 20:27getting started with the very first chaos engineering experiment
- 21:06augmenting chaos engineering with ai and gen ai
- 24:50demo: ai-drive chaos engineering platform
- 37:28conclusion & takeaways
- 38:22thank you
related talks
57mChaos Engineering Primer
38mChaos Engineering
18mFostering Chaos Engineering Culture
27mChaos Engineering: The Path to Reliability
01h 04mHarnessing Chaos; the hidden ingredient behind building better systems through learning
18mChaos Engineering: Break It On Purpose
34mChaos Engineering For Hybrid Targets With LitmusChaos
38mScientific and Safe Chaos Engineering