caching-strategies

Caching Strategies

When to use

• Designing cache layers for high-traffic systems
• Preventing cache stampede / thundering herd
• Setting CDN/HTTP cache headers or cache invalidation strategy

Core principles

1. Cache-aside is the default — application controls all cache interactions
2. Always add jitter to TTL — ±10% prevents synchronized expiry stampedes
3. Cache invalidation is harder than caching — design invalidation FIRST, then TTL
4. Multi-level cache reduces latency multiplicatively — L1 (in-memory) → L2 (Redis) → L3 (DB)
5. A cache miss should never cascade to DB overload — use lock-based prevention

References available

• references/caching-patterns.md — cache-aside, write-through, write-behind, multi-level with code
• references/redis-deep-dive.md — data structures for caching, eviction policies, cluster mode
• references/cdn-edge-caching.md — Cache-Control headers, stale-while-revalidate, purge strategies
• references/cache-invalidation.md — stampede prevention, event-driven invalidation, versioned keys

Assets available

• assets/caching-checklist.md — what to cache, TTL recommendations, monitoring queries