SOA lessons learnt (OR Microservices done better)
summary
Argues that most microservices mistakes replay SOA failures and recommends concrete patterns: use messaging (example: NServiceBus) and queues for state changes, avoid synchronous cross-service calls, separate read/write (CQRS/CQS), give each bounded context its own data store (no referential integrity), and accept eventual consistency. Shows that a fire-and-forget write can make 10,000 inserts in two seconds versus blocking REST, warns of real-world scale like thousands of endpoints, and urges componentized monoliths until scale demands distribution.
AI-written summary · about
chapters (25)
- 0:00Introduction
- 2:22My story
- 4:44What is coupling
- 6:29Aspects of coupling
- 9:47Ignoring the network
- 12:21Distributed computing fallacies
- 13:46Business logic fallacies
- 15:00Microservices vs monolith
- 18:00Initial design
- 19:31Vertical slicing
- 22:33Openclosed principle
- 23:44State changing operations
- 25:03Query separation
- 27:59Smallness
- 29:42Synchronous messaging
- 32:49Be explicit
- 33:34Update user
- 35:40Single responsibility
- 38:37Database
- 39:38Business Systems
- 41:39Synchronization
- 42:52Composite
- 43:44Monitoring
- 45:22Summary
- 50:39Deployment
related talks
52mSOA lessons learnt OR Microservices done better
54mThe Long Sad History of Microservices
38mMicroservices, a bittersweet symphony
23mWhy You Had a Bad Time With Microservices
50mEmbracing Microservices
40mSOD Microservices - Daniel Jones, EngineerBetter
01h 00mService Oriented Architecture for Robust & Scalable Systems
36mDo Proper Monoliths before You Do Microservices!
