Django Enterprise Caching Strategies: Redis & Multi-Tier Caching

No matter how optimized your SQL queries are, computing the same database calculations for thousands of simultaneous visitors wastes CPU cycles and spikes latency. Django's Caching Framework stores pre-rendered HTML fragments or raw Python objects in high-speed in-memory memory stores like Redis.

Configuring Redis Cache in settings.py

Install the modern django-redis client package:

pip install django-redis

# mywebsite/settings.py
CACHES = {
    "default": {
        "BACKEND": "django_redis.cache.RedisCache",
        "LOCATION": "redis://127.0.0.1:6379/1",
        "OPTIONS": {
            "CLIENT_CLASS": "django_redis.client.DefaultClient",
            "CONNECTION_POOL_KWARGS": {"max_connections": 100},
            "SOCKET_CONNECT_TIMEOUT": 5,
            "SOCKET_TIMEOUT": 5,
        },
        "KEY_PREFIX": "veewom_cache"
    }
}

1. Per-View Caching with @cache_page

from django.views.decorators.cache import cache_page

# Caches the entire rendered response in Redis for 15 minutes (900 seconds)
@cache_page(60 * 15)
def homepage_view(request):
    featured_articles = BlogPost.objects.published().featured()[:5]
    return render(request, 'home.html', {'articles': featured_articles})

2. Template Fragment Caching

{% load cache %}

<!-- Cache this expensive category sidebar widget for 1 hour (3600s) -->
{% cache 3600 sidebar_nav %}
    <ul class="cat-list">
        {% for cat in all_categories %}
            <li><a href="{{ cat.get_absolute_url }}">{{ cat.name }} ({{ cat.articles.count }})</a></li>
        {% endfor %}
    </ul>
{% endcache %}

3. Low-Level Cache API with Cache-Aside Pattern

from django.core.cache import cache

def get_platform_metrics():
    # Cache-Aside Pattern:
    # 1. Check if key exists in Redis
    metrics = cache.get('global_platform_metrics')
    if metrics is None:
        # 2. Cache miss -> perform heavy calculations
        metrics = {
            'total_students': User.objects.count(),
            'total_tutorials': BlogPost.objects.filter(is_published=True).count(),
        }
        # 3. Store in Redis for 30 minutes
        cache.set('global_platform_metrics', metrics, timeout=60 * 30)
        
    return metrics