My monolith was falling apart. One slow email service brought down the entire checkout. Then I learned about message queues—RabbitMQ specifically—and everything changed. How I went from tightly coupled HTTP calls to asynchronous messaging, and what broke (and what didn't) along the way.
We broke the monolith. We built independent pipelines. We created a custom Pub/Sub event bus. Our architecture was a masterpiece of Webpack 5 engineering. But a few months later, I had to ask: Did we sacrifice developer sanity at the altar of scale? A reflection on the true cost of microfrontends.
Microfrontends are great until they need to talk to each other. How do you pass data between two React apps that don't share a state tree? Redux? Context? No. We went back to basics and built a vanilla JavaScript Pub/Sub Event Bus. Here's why it works.
With our Webpack Module Federation setup, we gained a superpower: independent deployments. The Cart team could deploy on Tuesday, and the Product team on Friday. No coordination meetings. No monolithic release trains. Here's how we orchestrated our dynamic remotes.
I started with MySQL. I dabbled with SQLite. Then I tried PostgreSQL, and I understood why experienced developers get weirdly passionate about it. JSON columns, full-text search, window functions, CTEs, JSONB indexing—Postgres isn't just a database, it's a Swiss Army data platform. Here's why I switched and never looked back.
I dismissed SQLite as a 'toy database' for years. No server, no configuration, just a file. Then I learned that it runs on every smartphone, every browser, every Mac, every Python install, and handles more read queries per day than all other databases combined. I was very, very wrong about SQLite.
Before I understood databases, I stored everything in JSON files and localStorage. Then I discovered MySQL, and suddenly data had structure, relationships, and rules. Here's how I went from 'what's a JOIN?' to actually building things that persist.