Monolith Decomposition Patterns
summary
Practical patterns to break a monolith into services: put a proxy (nginx or similar) in front to strangler fig route endpoints and measure added latency (expect a few ms; test for bad networks), extract features behind interfaces and flip with feature flags using branch by abstraction, avoid services reading the monolith DB long term (or you create a distributed monolith), move data via APIs/events or CDC, run parallel comparisons in production, use DB refactoring tooling like Flyway and expect extra round‑trips and consistency tradeoffs.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 0:33Monolith Micro Services
- 3:20Monoliths
- 8:00Distributed Monolith
- 14:06Release Trains
- 17:07Monolith
- 23:37You need a reason
- 26:14Dont get full into those traps
- 29:45Movement of functionality
- 34:32Strangler pattern
- 35:15Branch by abstraction
- 38:30Parallel run patterns
- 45:32Database refactoring
- 46:16Database refactoring example
- 49:21Database safety net
related talks
42mQ&A - Monolith to Microservices
25mBreak up Monoliths and lay them to REST!
43mMonolith Carving
36mDecomposing a monolith - engineering and operational strategies we have gleaned from our experience
57mModularizing the Monolith
01h 52mMasterclass: Let's Cut the Monolith
43mMicroMonolith - Top anti-patterns of distributed systems
47mFrom Monolith to Microservices With the Strangler Pattern