Considering Migrating a Monolith to Microservices?
summary
Argues a decision method: weigh five dark energy forces (simplicity, team autonomy, fast pipeline, tech stacks, special resource/criticality) against five dark matter forces (local interactions, efficiency, ACID vs Saga pattern, runtime and design coupling) to pick which monolith parts become services. Recommends modularize and fix DevOps first, use the Strangler fig pattern for incremental extraction, and cites a real case where a machine learning service was split out to Python with faster deploys and clearer ownership.
AI-written summary · about
chapters (7)
- 0:00Start
- 20:38Fast deployment pipeline
- 21:25Optimizing a monolith's deployment pipeline
- 23:21Support multiple technology
- 33:49Minimize runtime coupling
- 34:48Minimize design time coupling
- 36:45Consequences of dark matter forces
related talks
42mQ&A - Monolith to Microservices
37mEvolutionary Architecture & Microservices
38mFrom Monolith to Microservices
32mSplitting the monolith
01h 03mFrom Monolith to Microservices: Story and Practice
43mMicroMonolith - Top anti-patterns of distributed systems
26mPros/Cons of Microservices
50mEmbracing Microservices
