Sync vs Async Python Web Servers

This project demonstrates the difference between synchronous (Flask + uWSGI) and asynchronous (FastAPI) Python web servers under heavy load, using Docker Compose for easy setup.

What it shows

• Sync server (Flask + uWSGI): Drops requests under high load due to listen queue limits, resulting in many 5XX errors and high latency.
• Async server (FastAPI): Handles the same load reliably, with lower latency and almost no failed requests.

How to run

Build and start all services:

docker compose up --build

Stop and remove all containers:

docker compose down

Project structure

• sync_web/ – Flask app with uWSGI (sync)
• async_web/ – FastAPI app (async)
• tester/ – Load generator

Summary

This setup helps visualize how async Python servers can handle high concurrency more gracefully than traditional sync WSGI servers, which may drop requests under pressure.

---

Inspired by real-world production issues with WSGI servers dropping requests under load, as described in the accompanying Medium post.