REST API Performance Optimization

# CHAPTER 28

REST API Performance Optimization #

1. Introduction #

2. Learning Objectives #

• Identify common bottlenecks that slow down PHP REST APIs.

• Optimize MySQL queries (Indexing and N+1 queries).

• Implement GZIP/Brotli compression to reduce JSON payload sizes.

• Understand and implement HTTP Caching headers (Cache-Control, ETag).

• Implement Server-Side caching using Redis/Memcached.

3. Beginner-Friendly Explanation #

1. 1. Database Optimization: If you ask the librarian for all books by "Smith", and the books aren't sorted, they have to check every single book in the building (Full Table Scan). If the books are indexed alphabetically, they find them instantly.

1. 2. Payload Compression: If you request 100 books, the librarian packs them tightly in a vacuum-sealed bag (GZIP) so they are easier to carry to your car.

1. 3. Caching: If 50 people ask the librarian for the same daily newspaper, the librarian keeps a copy on the front desk (Cache) instead of walking to the back room 50 times.

4. Real-World Examples #

• Caching: An e-commerce API has a /categories endpoint. Categories rarely change. By caching this data, the API can serve 10,000 requests per second with zero database lookups.

• Compression: A /users endpoint returns a 1MB JSON string. By enabling GZIP, that string is compressed to 100KB before being sent over the network, making it 10x faster to download on a mobile phone.

5. Detailed Code Examples #

<?php
// Endpoint: GET /api/categories

// 1. Tell the client to cache this response for 3600 seconds (1 hour)
header(&#039;Cache-Control: max-age=3600, public');

// 2. We can also use an ETag (a hash of the data)
$data = [
    ["id" => 1, "name" => "Electronics"],
    ["id" => 2, "name" => "Clothing"]
];
$jsonOutput = json_encode($data);
$eTag = md5($jsonOutput);

header("ETag: \"$eTag\"");

// 3. Check if the client sent an ETag in their request header
$clientETag = $_SERVER[&#039;HTTP_IF_NONE_MATCH'] ?? '';

// If the client's hash matches our current hash, the data HAS NOT CHANGED!
if ($clientETag === "\"$eTag\"") {
    // Return a 304 Not Modified. The server sends NO body data!
    http_response_code(304);
    exit;
}

// Otherwise, send the full data
echo $jsonOutput;
?>

6. Request/Response Examples #

Client Request 1: GET /categories *(Downloads 500KB of JSON)*

Client Request 2 (5 minutes later): GET /categories If-None-Match: "eTagHash123" Response: 304 Not Modified *(Downloads 0KB. The app instantly uses the data it saved 5 minutes ago!)*

7. HTTP Examples #

<IfModule mod_deflate.c>
  AddOutputFilterByType DEFLATE application/json
</IfModule>

8. JSON Examples #

Bad (100KB payload): [{"id": 1, "name": "John", "bio": "long text...", "createdat": "...", "avatar": "..."}]

• Database Indexes: This is the #1 fix for slow APIs. Ensure columns used in WHERE or ORDER BY clauses (like userid or created_at) have indexes in MySQL.

• Solve the N+1 Problem: If you fetch 50 articles, and then run a separate SQL query inside a loop to fetch the author for *each* article, you are running 51 queries! Use SQL JOINs to fetch all data in a single query.

• Use Redis for Data Caching: Instead of querying MySQL for the top 10 articles on every request, query MySQL once, save the JSON string in Redis (RAM), and serve subsequent requests directly from Redis. It is 100x faster.

10. Common Mistakes #

• Caching User Data: Be extremely careful with caching. Never use Cache-Control: public on an endpoint like /profile, or an intermediary proxy server might cache User A's private data and accidentally serve it to User B! Use Cache-Control: private for user-specific data.

• Over-fetching: Returning a massive JSON object with 50 keys when the frontend only displays 3 of them.

11. Mini Exercises #

1. 1. What does the HTTP Status Code 304 Not Modified mean for an API response?

12. Coding Challenges #

13. MCQs with Answers #

14. Interview Questions #

• Q: Explain how ETags and the 304 Not Modified status code improve mobile app performance.

• Q: What is the N+1 query problem, and how do you resolve it in a relational database?

• Q: When should you use Server-Side caching (Redis) vs Client-Side caching (HTTP Cache-Control)?

15. FAQs #

16. Summary #

17. Next Chapter Recommendation #