Writing about development and technology.
SSR pages showed blank screens for 3 seconds during data loading. Streaming SSR sends ready parts first, dramatically improving perceived speed.
Creating API routes, fetch calls, loading states... Too much code for one form. Server Actions dramatically simplified form handling.
Passing a function as props from a Server Component broke everything. Understanding the serialization boundary revealed RSC's true patterns.
Public APIs face unexpected traffic floods without proper protection. Rate limiting, API key management, and IP restrictions to protect your API.
Started with admin/user roles but requirements grew complex. When RBAC isn't enough, ABAC provides attribute-based fine-grained control.
With 3 services needing separate logins, SSO unified authentication. One login grants access to everything.
Password resets were half my support tickets. Passkeys eliminate passwords entirely, but implementation is more complex than expected.
Images were 80% of my page weight. Format conversion to WebP/AVIF and responsive images dramatically improved loading speed.
Making your app work offline requires understanding Service Worker caching strategies. Cache-first, network-first, and stale-while-revalidate explained.
Parsing a CSV file froze the entire UI. Web Workers run heavy computation in a separate thread, keeping the UI responsive.
Imported one lodash function but got the entire library in my bundle. Making tree shaking actually work requires understanding ES modules.
2MB bundle meant 5-second initial load. Lazy loading and code splitting load only what's needed, cutting it to 2 seconds.
SQLite is just for mobile apps? Turso and Litestream made it a serious option for production web services with edge replication.
Adding a column to a production table locked it for 5 minutes. Learning zero-downtime migration strategies the hard way.
You don't always need Elasticsearch. PostgreSQL's built-in Full-Text Search handles most search needs without extra infrastructure.
Beyond basic SELECT: JSONB, CTEs, and Window Functions dramatically expand what you can do in a single SQL query.
I only used Redis as a cache until I discovered it handles sessions, leaderboards, pub/sub, and rate limiting. It's a Swiss Army knife.
Manual builds and deploys led to shipping bugs to production. GitHub Actions automated my pipeline and eliminated human error.
Serving static files from your server works at small scale, but traffic spikes can bring it down. S3 + CloudFront separates static delivery from your app server for any scale.
I've deployed on all three platforms. Each has clear strengths in pricing, speed, and features. Here's how to choose for your project.
