Caching Strategies — ElastiCache, DAX, CloudFront Caching

Tổng quan

Caching giảm latency và load cho database/origin. AWS cung cấp nhiều lớp cache: edge (CloudFront), application (ElastiCache), database-specific (DAX). Chọn đúng layer và pattern là key.

ElastiCache Redis vs Memcached

• Redis: data structures, persistence, replication, Multi-AZ, pub/sub, Lua scripting, cluster mode.
• Memcached: simple key-value, multi-threaded, no persistence, no replication.
• Redis Cluster Mode: sharding data across nodes, scale write + read.
• Redis Global Datastore: cross-region replication cho DR.
• 99% trường hợp: chọn Redis. Memcached chỉ khi cần simple ephemeral cache.

Cache Patterns

• Lazy Loading (Cache-Aside): read miss → query DB → write cache. Stale data possible.
• Write-Through: write DB + write cache đồng thời. Always fresh, write penalty.
• Write-Behind: write cache → async write DB. Fast write, risk data loss.
• TTL: set expiration, balance freshness vs hit rate.
• Cache Invalidation: application-level delete/update cache khi data thay đổi.

DAX & CloudFront Caching

• DAX: in-memory cache cho DynamoDB, microsecond latency, API-compatible (drop-in).
• DAX cluster: primary + read replicas, Multi-AZ.
• DAX vs ElastiCache: DAX chỉ cho DynamoDB, ElastiCache cho mọi DB.
• CloudFront: edge cache cho HTTP content, TTL-based, cache key (headers, cookies, query strings).
• CloudFront invalidation: tạo invalidation request hoặc dùng versioned URLs.