How would you design a caching strategy for a high-traffic web application with stale data tolerance?

Multi-Layer Caching Architecture

Browser Cache → CDN Cache → Application Cache (Redis) → Database
  (client)      (edge)        (server)                   (source)

Layer 1: Browser Cache

// Static product data — cache aggressively
app.get("/api/products", (req, res) => {
  res.set("Cache-Control", "public, max-age=300, stale-while-revalidate=600");
  // Fresh for 5min, serve stale for up to 10min while revalidating
  res.json(products);
});

Layer 2: CDN Cache (Cloudflare/Fastly)

Rules:
- /api/products/* → Cache 5 min, serve stale 10 min
- /api/products/[id] → Cache 10 min, serve stale 15 min
- /api/search → No cache (dynamic)

// Surrogate-Key for targeted invalidation
res.set("Surrogate-Key", "products product-123");
// When product 123 updates: PURGE product-123 from CDN

Layer 3: Redis Application Cache

class CacheService {
  private redis: Redis;

  async get<T>(key: string, fetcher: () => Promise<T>, ttl = 300): Promise<T> {
    const cached = await this.redis.get(key);
    if (cached) {
      const parsed = JSON.parse(cached);

      // Stale-While-Revalidate pattern
      if (parsed.expiresAt < Date.now()) {
        // Stale — return it but refresh in background
        this.refreshInBackground(key, fetcher, ttl);
      }
      return parsed.data as T;
    }

    // Cache miss — fetch and store
    const data = await fetcher();
    await this.set(key, data, ttl);
    return data;
  }

  private async refreshInBackground<T>(key: string, fetcher: () => Promise<T>, ttl: number) {
    fetcher().then(data => this.set(key, data, ttl)).catch(console.error);
  }

  private async set<T>(key: string, data: T, ttl: number) {
    await this.redis.set(key, JSON.stringify({
      data,
      expiresAt: Date.now() + ttl * 1000,
    }), "EX", ttl * 2); // Redis TTL = 2x logical TTL
  }
}

Cache Invalidation Strategies

Strategy	How	Best For
TTL-based	Auto-expire after N seconds	Data with known freshness needs
Event-driven	Invalidate on write	User-specific data
Tag-based	Invalidate by group (CDN)	Related content
Write-through	Update cache on DB write	Frequently updated data

The 3 Golden Rules

1. Cache at the right layer — Browser for static, CDN for semi-static, Redis for dynamic
2. Use stale-while-revalidate — Never block users on cache refresh
3. Invalidate smartly — Tag-based for bulk, event-based for single items

Traffic Flow at Scale

50,000 req/sec incoming
  → 45,000 served from CDN (90%)
  → 4,500 served from Redis (9%)
  → 500 served from Database (1%)

Database load: 500 qps (well within 5K capacity) ✅