How to Fail at Building Modern Microservices
summary
A how-to list of concrete anti-patterns: keep a 30‑minute startup monolith and put the 30‑minute app in a Docker container; bake layers so every rebuild pushes a 150MB JVM layer; refuse contract testing and focus only on unit tests; organize single‑discipline teams and hard‑code URLs; deploy quarterly, not continuously; give each service its own Kubernetes cluster so utilization stays ~2% and costs explode; centralize no logs or tracing to preserve obscureability and reward release heroes.
AI-written summary · about
chapters (11)
- 0:00Introduction to failure
- 3:11Containerizing monoliths poorly
- 8:28Sabotaging team structure
- 12:03Tightly coupled services
- 15:29Ignoring distributed computing truths
- 17:41Exploiting testing gaps
- 20:14Quarterly delivery sabotage
- 22:02Avoiding automation
- 24:07Celebrating release heroes
- 25:15Maximizing cloud costs
- 26:38Obscuring system behavior
related talks
40m10 tips for failing at microservices
39m10 Tips for failing at microservices
57mEmbracing Microservices
55mMicroservices, Where Did It All Go Wrong
43mHow to Avoid Doing Microservices Completely Wrong
47mEmbracing Microservices
44mMicroservices, Where Did It All Go Wrong
53mDon’t Build a Distributed Monolith: How to Avoid Doing Microservices Wrong