Django Ninja Pydantic Benchmark

This repository contains benchmarking code to compare the performance of different schema validation approaches in Django Ninja and Pydantic.

Overview

This benchmark compares the performance of:

1. Django Ninja Schema: The standard Schema class from django-ninja
2. SimpleDjangoSchema: A simplified version of Django Schema with resolver logic removed
3. Pydantic BaseModel (v2): The current version of Pydantic's BaseModel
4. Pydantic BaseModel (v1): The previous version of Pydantic's BaseModel

The benchmark measures serialization/deserialization performance with different data sizes to help developers make informed decisions about which schema approach to use for optimal performance.

Key Findings

• Pydantic v2 (BaseModel) is significantly faster than both Pydantic v1 and Django Ninja Schema
• The performance difference becomes more pronounced with larger data sizes
• SimpleDjangoSchema offers improved performance over the standard Schema
• Visualizations clearly show the relative performance differences

Requirements

• Python 3.8+
• Django
• Django Ninja
• Pydantic (both v1 and v2)
• Matplotlib (for visualization)

How to Run

1. Clone the repository:

git clone https://github.com/BEMELON/django-ninja-pydantic-benchmark.git
cd django-ninja-pydantic-benchmark

2. Install the required dependencies:

pip install -r requirements.txt

3. Run the benchmark:

python benchmark_code.py

The script will run benchmarks for different sample sizes and display the results, including visualizations.

Customization

You can adjust the following parameters in the script:

• SAMPLE_SIZES: List of sample sizes to test (number of objects)
• ITERATIONS: Number of times to repeat each test for more reliable results

Visualization

The benchmark generates several visualizations:

1. Average execution time by sample size
2. Log-scale execution time for better comparison of values with different magnitudes
3. Performance multiplier bar graph showing how many times faster each implementation is

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests to improve the benchmarks or add more validation approaches.

Sample Output