Long-lived microservices
summary
Practical account of a long-lived microservices stack that actually ran in production: Scala services and some C++ components use Akka Persistence, Kafka journal as the event store and protobuf envelopes for interoperable messages, with JWT value-based auth, circuit breakers for offset storage failures, CQRS for push views and grouping via Akka HTTP pools. Code, Docker setups and property tests (RapidCheck) live on GitHub; approach reduces downtime at the cost of potential duplicate delivery.
AI-written summary · about
chapters (19)
- 0:00Intro
- 1:06Design & architecture
- 3:03Details make things complex
- 6:14Stateful services
- 7:19Message delivery semantics
- 10:32Details make the difference
- 10:43Good protocols
- 13:16Tooling: build-time support
- 14:38Tooling: "natural-looking" code
- 15:23Tooling: testing
- 16:10Strict and fully-contained
- 17:28Security is important
- 17:48Value semantics for authn & z
- 23:10Check claims in every service
- 25:00Reactive
- 28:48Resilient
- 35:01With back-pressure
- 38:38Assemble entire systems
- 41:38Summary
related talks
51mImplementing Microservices with Scale and Akka
59mImplementing Microservices with Scala and Akka
55mExperience report on using Akka in production: pitfalls and rewards
53mHow do we build high‑load microservices
45mReactive Microservices with Akka and Docker
31mHundreds of Microservices without breaking your APIs
48mChallenges in Implementing MicroServices
51mMicroservices for Mortals