Async vs Sync Benchmark

This project contains a small FastAPI application backed by a PostgreSQL database and a pair of benchmark clients used to compare synchronous and asynchronous HTTP requests. It can be used to observe latency and throughput differences when making many concurrent requests.

Results

Sync

Async

Requirements

• Python 3.13+
• PostgreSQL database

The project dependencies are declared in pyproject.toml. Install them with pip:

pip install -r uv.lock

Environment Variables

The database connection details are provided through environment variables:

• DATABASE_URL – connection string used by the application
• DB_NAME, DB_USER, DB_PASSWORD, DB_HOST, DB_PORT – used by the data population script

Create a .env file or export the variables in your shell before running the code.

Running the Application

Start the FastAPI server with uvicorn:

uvicorn main:app --reload

The service exposes a single endpoint:

• GET /clientes/{id} – returns a client record from the database

Populating Sample Data

Use the generator_fake_clients.py script to seed the database with sample clients:

python generator_fake_clients.py

Benchmark Clients

Two benchmark scripts are provided:

• benchmark_client_async.py – performs requests using httpx.AsyncClient
• benchmark_client_sync.py – performs requests using requests

Each script will issue 10,000 requests against the running API and print throughput and latency metrics.

Run them with:

python benchmark_client_async.py

python benchmark_client_sync.py

Compare the output numbers to evaluate the performance differences between asynchronous and synchronous requests.