The Magic of Small Things - 10 Years of Microservices
summary
Argument: microservices become the dominant style because splitting huge monoliths (five to twenty million lines, clients with 250 million lines of COBOL) into small, independently deployable components reduces coordination and speeds teams, but shifts complexity into infrastructure. Concrete tooling: Docker, Kubernetes, Kafka/RabbitMQ, gRPC, AWS Lambda, Chaos Monkey, CI/CD and observability stacks — platform engineering and infra-as-code mostly work but create “Lambda pinball” and a monitoring explosion. Decentralized patterns like decentralized data management and product teams improve outcomes but distributed transactions and governance remain the hardest problems; 2023 DevOps data shows measurable team performance gains.
AI-written summary · about
chapters (16)
- 0:00Intro
- 4:16Stuff that was on our mind
- 9:07(my version of the) Origin story
- 21:20Componentization via services
- 27:01Organized around business capabilities
- 29:47Decentralized data management
- 31:56Products no projects
- 35:23Decentralized governance
- 42:50Smart endpoints & dumb pipes
- 49:36Evolutionary design
- 52:24Infrastructure automation
- 56:30Designed for failure
- 58:15Unexpected side-effects
- 59:58What I wish we'd done differently
- 1:01:34Are microservices the future?
- 1:02:10Outro
related talks
44mIs It Time Already? Microservices Are Dead; Long Live Microservices!
45mMicroservices: A retrospective
01h 00mThe Magic of Small Things – Microservices 10 Years On
37mKnee Deep in Microservices
01h 18mIn Defense of Microservices
52mMicroservices: A Utopian Mystery
14mSAG Expert Interview #05: The Evolution of Microservices
55mMicroservices, Where Did It All Go Wrong