Why Node.js Needs an Application Server
summary
Claim: run Node with a dedicated control plane using worker threads so a coordinator observes blocked event loops externally while workers handle requests. Architecture uses Prometheus on the main thread and an internal mesh to run Next.js + Fastify + microservices in one process, support hot-swap workers without dropping traffic, restart crashed workers immediately, and scale workers independently. One customer reports halving infrastructure costs by condensing many microservices into a single deployment unit.
AI-written summary · about
chapters (14)
- 0:00Intro
- 0:57Your 20% Off 🤫
- 1:20Node.js Is Everywhere. And That's Exactly the Problem
- 2:16Wait, Node.js Is Still Single-Threaded? Where Have You Been?
- 3:28The Three Pillars of Running Node in Production Nobody Talks About
- 4:18The Monitoring Paradox: Your App Goes Blind Right When You Need It Most
- 6:25Meet Watt: The Architecture That Actually Fixes This
- 7:36How the Coordinator Keeps Watch While Your Workers Are Drowning
- 8:40Reliable Health Checks Under Heavy Load? Yes, It's Possible
- 9:39What Happens When a Worker Crashes - and Why It's Not a Big Deal Anymore
- 10:44Running Next.js + Fastify + Node in One Process: Live Demo
- 14:24Scaling Workers Up and Down in Real Time
- 17:21Blocking the Event Loop on Purpose - and Watching Nothing Break
- 18:20Q&A: Mixed CPU + I/O Workloads, Fault Isolation, PM2 Migration, HMR in Dev
related talks
46mArchitecture of a Modern Enterprise Node js Application
30mThe working Architecture of Node.js Applications
32mNODE.JS WORLD
22mMulti-tiered Node.js architectures
35mRethinking Enterprise Architectures with Node.js
35mNodeJS: from single thread to worker threads
39mYou Don’t Node.js
22mScaling Node.js and unblocking the Event Loop