Microservices: the small print
summary
Compares a Django monolith versus a simple microservices split for a task manager API (users, tasks, invoices with three JSON endpoints), demonstrating concrete code patterns: ORM joins and cascade delete inside one DB versus per-invoice HTTP fetches to compute a task’s total and multiple remote deletes for one operation. Shows failures from distributed transactions and practical mitigations like bulk endpoints, retries and background repair leading to eventual consistency. Gives real replica counts for monoliths (Wikipedia, Reddit) and argues higher infra cost and network latency for microservices.
AI-written summary · about
chapters (12)
- 0:00Introduction to microservices
- 1:49Monolith vs microservices setup
- 2:57Adding task fee calculation
- 5:45Network calls and latency
- 6:50Service boundary design
- 8:16Transactional operations
- 10:42CAP theorem trade-offs
- 12:49Scaling monoliths
- 13:51Scaling microservices
- 15:38Conway's Law and teams
- 16:00Microservices small print
- 17:27When to avoid microservices
related talks
26mPros/Cons of Microservices
54mThe Long Sad History of Microservices
43mMicroMonolith - Top anti-patterns of distributed systems
55mMicroservices, Where Did It All Go Wrong
01h 03mMicroservices vs The Distributed Monolith
01h 00mMicroservices, Where Did It All Go Wrong?
19mMicroservices vs. Monolith for Scalability and Efficiency
39mConsidering Migrating a Monolith to Microservices?