Design a URL Shortener (TinyURL): Interactive System Design
URL Shortener
Guided ModeStep 4 β Cache for Fast Redirects
URL lookups (GET /:code) are read-heavy. The same popular links get clicked thousands of times. That's identical DB queries, each round-trip.
We introduce Redis Cache between the API and the Database.
Request Flow
Request: GET /abc123
The Load Balancer routes the redirect request to an API Server.
API checks Cache first
Before touching the Database, the API checks Redis Cache: "Do you have the long URL for code abc123?"
Cache HIT β 95% of traffic
Code found in cache β return the long URL instantly. Redis Cache responds in <1ms. Database is never touched.
// Cache hit path (95% of requests)
redis.get('url:abc123') // β 'https://longurl.com'
// Total latency: ~5msCache MISS β first visit
Code not in cache β query Database β store result in Redis Cache β redirect. Cache is populated; next visit is a hit.
// Cache miss path (5% of requests)
redis.get('url:abc123') // β null
db.query('SELECT...') // β 'https://longurl.com'
redis.set('url:abc123', 'https://...', 'EX', 86400) // TTL: 24hWhy We Need This
URL mappings almost never change. Once created, they're read thousands of times with the same result each time β the perfect cache target.
With 95% cache hit rate: the Database only sees 5% of real read traffic. You effectively get 20Γ DB capacity for free.
Key Insight
With a 95% cache hit rate, your Database only handles 5% of redirect queries. The same hardware now serves 20Γ more traffic β no new servers needed.
Step 4 of 6
