High-Speed URL Shortener

Overview

A high-throughput URL shortener I designed and deployed on Kubernetes, built around one question: how cheap can the redirect path get. Everything else in the system exists to keep that path fast.

Role

I designed the API, the persistence model, and the caching strategy, implemented the service in Go, and owned the deployment — horizontal autoscaling, load balancing, and the CI/CD pipeline that shipped it.

Problem

Redirect traffic is overwhelmingly reads, and every millisecond shows. Hitting the database on each request is trivial to build but buckles under bursty load — and blindly trusting destination URLs invites abuse.

Solution

PostgreSQL holds the canonical URL records; Redis serves the hot redirect mappings. Short codes come from a collision-checked generator, redirects validate their destination before responding, and analytics writes stay off the critical path so tracking never slows a redirect. API rate limiting caps abuse at the edge.

Outcome

A cache-first read path with a clean fallback when Redis misses, running on Kubernetes with horizontal autoscaling behind a load balancer, and a CI/CD pipeline that deploys it without manual steps.