Monoliths strikes back...
summary
Argues that real decisions come from concrete tradeoffs: Amazon Prime Video consolidates its microservices onto a single EC2/ECS setup and reports a >90% cost reduction after finding microservices only drove a ~5% scaling benefit; recommends starting with a modular monolith (Django) using htmx, Deploy on render, enforce CI with GitHub action, offload async jobs with Celery + Redis, and migrate to services only when measurable load or cost justifies it.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:56Monolithic architecture
- 6:04Microservices promise and pitfalls
- 9:59Amazon Prime Video failure
- 12:55Monoliths vs microservices
- 15:29Modular monoliths
- 19:32Tools for modern monoliths
- 23:06Monolithic development guidelines
- 25:57Microservices pitfalls in startups
- 26:41Q&A
related talks
37mModern HTTP Architecture
34mThe Internet is your Application Blueprint
33mMicro-frontends: a migration story on the Edge
22mDEVOPS DAYS MTL 2018 - Alexandre Gervais
12mHow a micro-frontend approach works on a large scale web application
30mEvolving Serverless Architectures
28mOn a shoe-string and a t2.small: scaling on a [zero] budget.
32mLunch Panel