Growing & Thriving in a Multi-model World
summary
Argues that purpose-driven decomposition and visual maps prevent the Big Ball of Mud: a monolith with 10–12 bounded contexts stays with a flat cost of evolution for 10 years using CQS and event sourcing, while discipline (never altering a working box) and tools like context maps and ADR capture make splits affordable. Recommends three alternatives per decision, prototype risky areas, map fear and dependencies, and even use LLMs for early foreign-market research to reduce uncertainty and start splitting when signals are cheap.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:44Part 1: Good old times
- 6:10Summary: Good old times
- 7:14Starting with the right foot
- 8:11Bounded context
- 11:04Discipline isn't free
- 26:16Running faster
- 28:44When to split?
- 34:39Detecting where to start
- 37:51Multi-country
- 44:03Poking the landscape
- 45:02Summary
- 47:37Outro
related talks
55mCharting the uncharted: Discovering and evolving bounded contexts
01h 01mBounded Contexts: Manage the Understandability of Your Systems
37mBuilt to last: A domain-driven approach to beautiful systems
41mSystems Thinking by combining Team Topologies with Context Maps
41mUsing Boundary-Driven Development to beat code complexity
43mGood Fences Make Good Neighbours
50mTowards an Evolutionary Architecture
52mEvolving a Pragmatic, Clean Architecture A Craftsman's Guide