Simulating the Chaos of Real-World Networks for Resilient Applications
summary
An in-house test rig simulates real-world network chaos using a cheap OpenWRT router, a Flask UI and shell scripts that call tc traffic control, proving reproducible scenarios like urban canyon, high-speed driving and peak-hour congestion. Prototype costs ~€100 and takes a few weeks to build. It reveals concrete failures (lossless streaming stutters because device buffers are too small), drives fixes (adaptive buffering) and validates recovery behavior before rollout.
AI-written summary · about
chapters (32)
- 0:00Welcome
- 1:50Speaker and problem
- 3:14Network instability impact
- 5:22Why bad networks
- 6:52Simulating chaos essential
- 7:33In-house impairment tool
- 8:15Tool architecture
- 10:00Network simulation engine
- 11:24Tool UI demo
- 12:26Impairments and relevance
- 14:36Scenarios vs ad hoc
- 15:41AI for scenarios
- 16:25Alternative tools
- 18:55Lossless streaming need
- 20:01Streaming under variable
- 21:47Key takeaways
- 22:49Q&A
- 23:14Network performance issues
- 24:37Open-sourcing tool
- 25:18Fixing streaming issue
- 26:24VPN testing scenarios
- 27:04Why not out-of-box
- 28:29Automated pipelines
- 29:32Device selection
- 30:14Tool compatibility
- 31:18Open QA roles
- 32:01App stabilization
- 32:22Windows alternatives
- 33:02Tool resources
- 34:26iOS offline best
- 35:53Charles proxy scripts
- 36:13Closing
related talks
06m"Outperforming in the Chaotic World!"
23mHow to tame chaos effectively?
01h 02mPractical Chaos Engineering breaking things on purpose to make them more resilient against failure
38mScientific and Safe Chaos Engineering
38mChaos Engr. in the Fast Lane
15mConfidence in Chaos
45mDeveloping a Chaos Architecture Mindset
29mPractical distributed systems: Building for the real world with chaos engineering