1.3.8 Data Caching Services

Data Caching Services

So sánh Caching Services

Caching Strategies

Lazy Loading (Cache-Aside)

App → Cache (miss) → DB → Cache (write) → App
App → Cache (hit) → App

• ✅ Chỉ cache data được request
• ✅ Node failure không critical
• ❌ Cache miss = 3 round trips (penalty)
• ❌ Stale data possible

Write-Through

App → Cache (write) → DB (write)
App → Cache (read, always hit)

• ✅ Data luôn fresh
• ❌ Write penalty (2 writes mỗi lần)
• ❌ Cache chứa data có thể không bao giờ đọc

TTL (Time to Live)

• Kết hợp với Lazy Loading hoặc Write-Through
• Expire cached data sau time period
• Balance giữa freshness và performance

DynamoDB DAX

App → DAX Cluster → DynamoDB
     (microsecond)   (millisecond)

• In-memory cache cho DynamoDB
• Read: GetItem, BatchGetItem, Query, Scan
• Write: Write-through (DAX → DynamoDB)
• API compatible với DynamoDB (chỉ đổi endpoint)
• Item cache + Query cache

import amazondax

# Chỉ cần đổi client
dax_client = amazondax.AmazonDaxClient(
    endpoints=['dax-cluster.abc123.dax-clusters.us-east-1.amazonaws.com:8111']
)
response = dax_client.get_item(
    TableName='Orders',
    Key={'id': {'S': 'order-123'}}
)

ElastiCache Redis Use Cases