Skip to content

Case Study: URL Shortener

A URL shortener (like bit.ly) maps short URLs to long URLs. Core: hash/encode long URL → short ID, store mapping in DB, redirect on access. Scale: ~100M URLs/month, read-heavy (100:1 read/write). Use Base62 encoding or counter-based ID, cache hot URLs in Redis, replicate DB for reads. Key decisions: custom aliases, analytics, TTL/expiration.

System Design

Step 1: Requirements

Functional: - Given a long URL, generate a short URL - Redirect short URL → long URL - Optional: custom aliases, expiration, analytics

Non-Functional: - Low latency redirects (< 50ms) - Highly available (redirects must always work) - Short URLs should be hard to guess

Estimation (100M new URLs/month): - Write: ~40 URLs/sec - Read: ~4000 redirects/sec (100:1 ratio) - Storage: 100M × 500 bytes ≈ 50 GB/year

Step 2: High-Level Design 
Client → API Gateway → URL Service → Database
                           ↓
                        Cache (Redis)

API: 

POST /api/shorten  { "longUrl": "https://..." }  → { "shortUrl": "https://short/abc123" }
GET  /abc123       → 301 Redirect to long URL

Step 3: Short URL Generation

Approach 1: Base62 Encoding of Auto-Increment ID 

ID: 12345 → Base62: "3dE"
Character set: [a-z, A-Z, 0-9] = 62 chars
6 chars → 62^6 = 56 billion combinations
7 chars → 62^7 = 3.5 trillion combinations

Approach 2: Hash + Truncate 

MD5("https://example.com/long/path") → first 7 chars
Problem: collisions. Check DB, retry if collision.

Approach 3: Pre-generated Key Service 

Key Service generates unique keys in advance
URL Service requests a batch of keys
No collision checking needed

Step 4: Database Design 
CREATE TABLE url_mappings (
    id BIGINT PRIMARY KEY,
    short_code VARCHAR(7) UNIQUE NOT NULL,
    long_url TEXT NOT NULL,
    created_at TIMESTAMP DEFAULT NOW(),
    expires_at TIMESTAMP,
    click_count BIGINT DEFAULT 0
);
CREATE INDEX idx_short_code ON url_mappings (short_code);

Read-heavy → cache hot URLs in Redis: 

Key: "url:abc123"
Value: "https://example.com/very/long/url"
TTL: 24 hours

Step 5: Scaling
  • Cache: Redis for frequently accessed URLs (80/20 rule)
  • DB replication: Read replicas for redirect lookups
  • DB sharding: Shard by hash of short code
  • CDN: Cache 301 redirects at edge
  • Rate limiting: Prevent abuse of URL creation
Common Follow-up Questions
  • How do you handle collisions?
  • How do you implement analytics (click tracking)?
  • How do you handle URL expiration?
  • What HTTP status code: 301 (permanent) vs 302 (temporary)?
  • How would you prevent abuse?