System Design: URL Shortener

The Problem

Design a URL shortening service like bit.ly. It should generate short URLs, redirect to the original, and track click analytics. Scale: 100M URLs created/month, 10B redirects/month.

API

POST /api/shorten  { "url": "https://..." }  → { "short": "https://short.ly/abc123" }
GET  /abc123       → 301 redirect to original URL
GET  /api/analytics/abc123  → { "clicks": 12345, "by_country": {...} }

Key Design Decisions

Short Code Generation

Base62 encoding of an auto-incrementing counter: 0-9, a-z, A-Z → 62^7 = 3.5 trillion unique codes

Pre-generate codes in batches to avoid DB bottleneck

Check uniqueness via database unique constraint

Storage

Postgres for URL mappings (short_code → original_url, created_at, user_id)

Redis for caching hot URLs (90% of redirects hit 10% of URLs)

Redirect Flow

Request → Load Balancer → App Server → Redis cache check
                                          ↓ miss
                                       Postgres lookup → cache result → 301 redirect

Analytics

Log clicks to Kafka → aggregate hourly in a batch job → store in analytics table

Don't query the click log directly (too large)

Scale Numbers

100M URLs/month = ~40 URL creates/second

10B redirects/month = ~4000 redirects/second → Redis easily handles this