Database Connection Pooling

Introduction

Connection pooling reuses database connections instead of creating new ones for each request, significantly improving performance.

Key Concepts

Connection Pool: Cache of database connections.

CONN_MAX_AGE: Django's built-in connection persistence.

Deep Dive

Django's CONN_MAX_AGE

python
# settings.py
DATABASES = {
    'default': {
        'ENGINE': 'django.db.backends.postgresql',
        'NAME': 'mydb',
        'CONN_MAX_AGE': 60,  # Reuse connections for 60 seconds
        # 0 = close after each request (default)
        # None = unlimited persistence
    }
}

Using django-db-connection-pool

python
# pip install django-db-connection-pool

DATABASES = {
    'default': {
        'ENGINE': 'dj_db_conn_pool.backends.postgresql',
        'NAME': 'mydb',
        'POOL_OPTIONS': {
            'POOL_SIZE': 10,
            'MAX_OVERFLOW': 10,
            'RECYCLE': 300,
        }
    }
}

PgBouncer (External Pooler)

ini
# pgbouncer.ini
[databases]
mydb = host=localhost dbname=mydb

[pgbouncer]
pool_mode = transaction
max_client_conn = 100
default_pool_size = 20

Best Practices

1. Use CONN_MAX_AGE > 0 in production: Reduces connection overhead.
2. Match pool size to workers: Gunicorn workers × connections per worker.
3. Use transaction pooling for Django: Compatible with ORM.

Summary

Connection pooling improves performance by reusing database connections. Start with CONN_MAX_AGE, consider PgBouncer for larger deployments.