Without Resilience, Nothing Else Matters
summary
Argument: production uptime beats architecture accolades, so resilience is a design requirement from day one, not an add‑on. Concrete practices: implement crash-only software and recursive restartability with supervisors that restart failing components, isolate failures with bulkhead compartments and an error kernel, use asynchronous, eventually consistent protocols (batchable, commutative, idempotent), and validate by chaos testing with Netflix’s Simian Army and multi‑JVM/network fault tests. These approaches reduce cascading failures and enable fast recovery.
AI-written summary · about
chapters (24)
- 0:00Introduction
- 2:15What is resilience
- 4:58What is complex
- 13:41Design for failure
- 13:59Social systems resilience
- 17:51meerkats
- 21:07Resilience is nothing
- 22:49How to manage failure
- 23:57Root cause
- 24:22Embrace crash
- 26:33Recursive wrist
- 27:21State management
- 32:51Failure management
- 34:05Cascading failure
- 35:40Supervision
- 36:15Selfheal
- 39:37Error Kernel
- 44:30Diversity and redundancy
- 45:25Distributed systems
- 46:20Life beyond the illusion of present
- 46:40Decoupling in time
- 47:24Resilient protocols
- 49:21Testing Tools
- 50:12Summary
related talks
22mResilience Is Not Enough
01h 05mKeynote - Without Resilience, Nothing Else Matters
16mResilient by Design: Architectures for Crisis Response & Recovery
16mEngineering Failure-Resilient Systems
43mResilience Engineering Mythbusting
10mDesigning for Failure using AWS: Resilient Systems
58mResilient by Design
11mResilience engineering mythbusting