DevOps in the Cloud: Case Studies of Amazon.com teams and their resilient architectures
summary
Amazon runs two-pizza teams that own services end-to-end; a product detail page stitches calls from hundreds of backend services with a hard rule of graceful degradation (noncritical widgets must add zero latency). Search serves ~84,000 req/s at peak and a search resilience team builds a chaos orchestrator around AWS Fault Injection Service with SLO guard rails, scheduler and an Andon stop button. The Relay trucking app adopts a multi-region active-active architecture using Lambda and DynamoDB Global Tables so critical modules can fail over.
AI-written summary · about
chapters (7)
- 0:00DevOps culture at Amazon
- 4:34Two Pizza teams
- 9:27Detail page framework
- 12:18Chaos engineering
- 18:01Amazon Relay resilience
- 25:25Internal tools for teams
- 26:08Q&A
