Intro

title: "Introduction to API Caching — Why Caching Matters" description: "API caching stores frequently requested responses in a fast-access cache layer, reducing latency by 80-95% and cutting server load by reducing redundant processing." date: 2026-06-28 lastmod: 2026-06-28 weight: 11 tags: [apis, caching] }

API caching stores frequently requested responses in a fast-access cache layer, reducing latency by 80-95% and cutting server load by eliminating redundant processing.

What You'll Learn

• What API caching is and why it matters
• Cache hit vs cache miss
• Cache layers and topologies

Why It Matters

Without caching, every request hits your database and application servers. Caching reduces response times from 200ms to under 10ms and reduces server costs by up to 90%.

Cache Flow

flowchart LR
    C[Client] --> R[Reverse Proxy Cache]
    R -->|Cache Hit| C
    R -->|Cache Miss| S[App Server]
    S --> D[(Database)]
    D --> S
    S --> R
    R --> C

Code Examples

# Naive uncached response
@app.route('/products')
def get_products():
    products = db.execute("SELECT * FROM products")
    return jsonify(products)
    # Response time: 150-300ms

# Simple in-memory cache
cache = {}

@app.route('/products')
def get_products():
    if 'products' in cache:
        return cache['products']  # Response time: < 5ms
    products = db.execute("SELECT * FROM products")
    cache['products'] = jsonify(products)
    return cache['products']

# Redis cache layer
import redis
r = redis.Redis(host='localhost', port=6379, decode_responses=True)

@app.route('/products')
def get_products():
    cached = r.get('products')
    if cached:
        return jsonify(json.loads(cached))
    products = db.execute("SELECT * FROM products")
    r.setex('products', 300, json.dumps(products))
    return jsonify(products)

Common Mistakes

1. No Cache at All

Every request hits the database, causing unnecessary load.

2. Caching Everything

User-specific responses, mutations, and real-time data should not be cached.

3. No Cache Invalidation

Stale data persists in cache, serving outdated information.

4. Cache Poisoning

Unvalidated cached data can serve malicious content.

5. Ignoring Cache Headers

HTTP cache headers let intermediate proxies cache responses.

Practice Questions

1. What is a cache hit?
2. What is the typical latency improvement from caching?
3. What types of data should not be cached?
4. What is a reverse proxy cache?
5. How does caching reduce server costs?

Answers:

1. A request served from cache without hitting the origin server.
2. 80-95% reduction, from hundreds of milliseconds to single digits.
3. User-specific data, authentication responses, and live data.
4. A server (like Nginx or Varnish) that sits in front of app servers and caches responses.
5. Fewer requests reach application servers, reducing compute and database costs.

Challenge: Measure uncached response times for an API endpoint. Implement an in-memory cache and measure the improvement.

FAQ