Once Upon a Vulnerability: Building Resilience in a Chaotic World
summary
Explains failure domains and how to design resilient distributed systems using concrete practices from Datadog: identify weak points, partition failures, and apply abstractions and separation of concerns. Covers evolving a system from single-site to global scale and defensive architectures that contain chaos so services degrade gracefully rather than fail catastrophically.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:10Failure domains
- 2:32Migrating to microservices
- 4:37Network domain failures
- 7:28Network failure tools
- 8:542020 incident and normalization
- 12:09Cloud zone resilience
- 17:27Control vs data plane isolation
- 18:50Kubernetes per zone
- 20:35Load spikes with Bulkhead
- 22:20Multi-region challenges
- 26:12Holistic failure solutions
- 26:54Chaos engineering
- 28:20SLI and SLO contracts
- 30:06Post-incident best practices
- 31:55Deviance normalization and risk
- 34:01Team organization for incidents
- 36:27Cloud elasticity myth
- 39:18Conclusion
related talks
55mGoing Resilient
22mResilience Is Not Enough
03mDeveloping a Chaos Architecture Mindset
51mPatterns for building resilient software systems
01h 03mPatterns for Resilient Architecture
53mResiliency and Availability Design Patterns for the Cloud
29mThe art of being resilient. How to handle failures gracefully in Distributed Systems.
14mAntifragility or taming the hydra