Autonomous microservices don't share data
summary
Claims that microservices must never share business data and that tech like .NET Core Docker kubernetes or serverless does not remove coupling; cites a horror case of 30,000 HTTP requests from RPC chains and Netflix-style overcall. Recommends event-driven publish/subscribe with tiny events — e.g. order accepted with just an order ID — separate databases per logical service, services as technical authorities, and asynchronous messaging so invoicing, shipping, etc. operate autonomously and avoid distributed monolith failure.
AI-written summary · about
chapters (20)
- 0:00Intro
- 0:26History lesson
- 4:32Object oriented programming
- 14:45microservices
- 15:25Netflix example
- 17:27Myths about microservices
- 19:16Json vs XML
- 20:41Remote procedure calls
- 21:14Example
- 22:05Losing autonomy
- 26:05Docker
- 33:55Design and Model
- 34:51Monolith
- 37:39Publishing
- 38:13Invoice
- 38:26Address
- 38:54Shipping
- 39:25Customer preferred
- 39:51Rightsized services
- 42:08Wrapup
related talks
53mAutonomous microservices don't share data. Period.
47mAutonomous Microservices Don't Share Data. Period
50mEmbracing Microservices
47mEvent-Driven Microservices = Microservices Done Right
57mEmbracing Microservices
01h 00mPrinciples for Building MicroServices
01h 04mDon’t Build a Distributed Monolith
01h 03mMicroservices vs The Distributed Monolith