Measuring Micro-services
summary
Argues that risk is reducible by measuring emergent behavior rather than CPU: track message flow rates and invariant ratios (e.g., add-item vs tax messages should be 1:1) and use partial deployments to validate changes (example: five login services, 20% routed to new version). Uses formal specs like TLA+ invariants (Amazon used ~700-line spec and found a bug by exploring 35 steps), and practical tooling such as InfluxDB / Kafka / New Relic plus the Seneca Node.js framework.
AI-written summary · about
chapters (7)
- 0:00Introduction and health analogy
- 2:53What are micro-services?
- 10:05Message flow patterns
- 13:40Risk reduction with micro-services
- 20:08Measuring systems in production
- 24:32Formal methods and invariants
- 33:04Conclusion and future steps
related talks
30mMeasuring Microservices
51mResponsible Microservices
36mA practical tale of assessing a Microservices Architecture
46mSurviving microservices
45mMicroservices: A retrospective
38mMicroservices, a bittersweet symphony
48mChallenges in Implementing MicroServices
31mHundreds of Microservices without breaking your APIs
