The Evolution From Monolith to Microservices to Serverless and Beyond
summary
Argues the architecture trend moves from coarse monoliths toward ever finer units because that increases innovation velocity, citing HP’s ~50k servers at roughly 10% utilization as motivation to avoid wasted capacity. Recommends gradual monolith decomposition, microservices owning their schema, and replacing RPC with publish/subscribe decoupling to limit blast radius. Calls out AWS Lambda, Kubernetes and Akka as tooling, and highlights emerging platforms CloudState / Stateful Functions and Dapr that enable event-sourced stateful functions which keep an entity in memory, chain functions by events, and cut friction and I/O costs.
AI-written summary · about
chapters (21)
- 0:00Introduction
- 0:43Visual Representation
- 2:00Why are we doing this
- 3:30Characteristics
- 4:05Monolith
- 7:10Landscape of Systems
- 8:25Micro Services
- 10:31Service
- 11:58Spikes
- 12:38HP Maiden Servers
- 13:43Virtual Machines
- 14:23Kubernetes
- 15:55Less friction
- 18:16Breaking up monoliths
- 18:42Tight coupling
- 20:54Microservice coupling
- 22:21Remote procedure calls
- 24:03An alternative
- 28:08Getting into service
- 33:22Event flow
- 38:54Stateless functions
related talks
01h 00mMicroservices and Serverless
41mYou Don’t Want to Build a Serverless Monolith
39mBuilding Highly Scalable Microservices with Serverless
14mThinking Twice About Migrating to Serverless
24mPatterns of Evolution from Monolith to Microservices
35mChopping the monolith
31mThe Evolution of Serverless: Past, Present and Future
37mEvolutionary Architecture & Microservices