Web Framework Performance Benchmark

This project compares the performance of different web frameworks and ORM combinations for a simple user count endpoint. The benchmark tests include popular Python frameworks (Django, FastAPI, Robyn) and Go's Gin framework, each paired with different database ORMs.

🏃 Tested Combinations

• Django + Django ORM
• Robyn + Pony ORM
• FastAPI + SQLAlchemy
• Robyn + Tortoise ORM
• Go + Gin + pgxpool

📊 Benchmark Results

Framework + ORM	Requests/sec
Go + Gin + pgxpool	23,085.57
Robyn + Tortoise	8,044.60
FastAPI + SQLAlchemy	1,511.57
Robyn + Pony	498.45
Django + Django ORM	430.60

🛠 Project Structure

• django_app/: Django application with Django ORM
• fastapi_app/: FastAPI application with SQLAlchemy
• robyn_app/: Robyn application with Pony ORM
• robyn_tortoise_app/: Robyn application with Tortoise ORM
• go_gin_app/: Go Gin application with pgxpool
• wrk_benchmark/: Benchmark runner using wrk

🚀 Setup & Running

1. Prerequisites:

   • Docker and Docker Compose
   • Make (for convenience commands)

2. Environment:

   • PostgreSQL database
   • Each application runs in its own container
   • Consistent test environment using Docker Compose
   • The database has a table called user with 1000 users in it.

3. Quick Start

   make install_dependencies
   make benchmark

4. Optional: Running benchmarks:

   # Rebuild and start specific services
   make rebuild_django
   make rebuild_fastapi
   make rebuild_robyn
   make rebuild_robyn_tortoise
   make rebuild_go_gin
   
   # Run individual benchmarks
   make django
   make fastapi
   make robyn
   make robyn_tortoise
   make gogini

🔍 Benchmark Configuration

• Load Test Parameters:
  • 4 threads (-t4)
  • 100 concurrent connections (-c100)
  • 10 second duration (-d10s)
  • Endpoint tested: /users (returns user count)

📝 Notes

• All services connect to the same PostgreSQL database
• Each framework is configured with minimal middleware and optimized for performance
• The benchmark measures a simple read operation (counting users)
• Results may vary based on hardware and specific deployment configurations

🤝 Contributing

Feel free to submit issues and enhancement requests!

📜 License

The MIT License (MIT)