DjangoCon Europe 2023 | Squeezing Django performance for 14.9 million users on WhatsApp

Django Testing Security

"Boost Django's performance for massive user bases, just like WhatsApp, with expert tips for debugging, optimization, and testing, ensuring fast and secure applications with over 14.9 million users."

Key takeaways

Use estimating one in 10,000 users saved to measure performance impact
Debug Toolbar or connection.queries can help identify slow endpoints
Be conservative with estimates, even with relatively small changes
Use Faker to generate realistic fake data for testing
Use a Postgres configuration option to show slow queries
Use Django’s built-in context_processors, like static/media and csrf, to improve performance
Create custom pagination class to improve performance
Set a timeout on SQL queries to improve performance
Use explain analyze to analyze slow queries
Test on a separate setup instead of production environment
Use log min duration statement to monitor slow queries
Be cautious of security when integrating with WhatsApp and Facebook
Use Django’s built-in query caching for performance gains
Consider using PyPy to get a performance boost
Use approximate counts instead of full counts to improve performance
Use cursor pagination for large datasets
Test with fake users and large traffic volumes
Use statistics on the database to improve query performance
Use a database that can handle large traffic volumes
Use a consultation service to develop and update applications

DjangoCon Europe 2023 | Squeezing Django performance for 14.9 million users on WhatsApp

More talks