Episode 3 – Dev Diary: Building TinyURL in Production
In this episode, I walk through the full journey of building and shipping a TinyURL-like service to a real production environment.
This is not a theory-heavy episode. It is a hands-on breakdown of what actually happens when you take a backend system from local development to production: domains, SSL, reverse proxies, caching, databases, CI/CD pipelines, and all the mistakes along the way.
I start with a simple idea - shorten a URL - and end up deep inside system design, distributed systems, and production debugging.
What I Talk About
Goals behind the project - Learning how to ship a real system to production - Deploying on AWS EC2 - Configuring domains and SSL - Understanding ACID guarantees in practice - Building a real CI/CD pipeline
What a URL shortener really does - Why URL shorteners are read-heavy systems - Base62 encoding and ID generation - Hashing vs sequential IDs - Collision avoidance and scalability tradeoffs - Redirects, caching, and performance
System architecture - FastAPI as the backend API - PostgreSQL for persistence - Redis for caching hot paths - Nginx as a reverse proxy and SSL terminator - Gunicorn for production-grade serving
Performance and scale - Cache-aside strategy with Redis - Why redirects must be fast - Read-heavy traffic patterns - Database bottlenecks and replication concepts
CI/CD and deployment - Automatic deployment on push to main - Docker and Docker Compose - Common pipeline mistakes - Environment configuration issues - Debugging production failures
What went wrong - Path and volume issues - Nginx misconfigurations - SSL and Certbot challenges - Redis connection problems - Async testing pitfalls
What I learned - How production systems actually fail - Why observability matters - How small misconfigurations cause big outages - How much caching changes system behavior - Why shipping teaches faster than tutorials
Technologies Used
- FastAPI – backend API
- Gunicorn – production server
- PostgreSQL – relational database
- Redis – caching layer
- Nginx – reverse proxy and SSL
- Docker & Docker Compose
- AWS EC2
- GitHub Actions – CI/CD
- Pytest – testing
What I Want to Do Next
- Improve storage efficiency in production
- Handle multiple containers safely for ID generation
- Add rate limiting and abuse protection
- Add analytics and observability
- Explore distributed ID generation strategies
References & Further Reading
- FastAPI: https://fastapi.tiangolo.com/
- Redis: https://redis.io/docs/latest/
- Nginx: https://nginx.org/en/docs/
- Gunicorn: https://docs.gunicorn.org/en/stable/
- PostgreSQL: https://www.postgresql.org/docs/
- Base62: https://en.wikipedia.org/wiki/Base62
Related articles: - Demystifying Nginx: https://medium.com/gitconnected/demystifying-nginx-from-reverse-proxy-to-secure-gateway-149c75b99ad8 - Shipping FastAPI like a Pro: https://medium.com/python-in-plain-english/shipping-fastapi-like-a-pro-my-ci-cd-pipeline-for-tinyurl-c27b68434f04
This episode is for anyone who wants to stop building demos and start shipping real systems — and is ready to learn from things breaking along the way.
