- Fully managed, highly available, in-memory cache for DynamoDB.
- Designed to reduce read congestion and deliver microsecond latency for reads.
- Ideal for read-heavy workloads.
Key Features
- Seamless integration with existing DynamoDB APIs — no changes to application logic required.
- Default Time to Live (TTL) for cache entries is 5 minutes (configurable).
- Caches individual items and query/scan results.
- Removes the need to manually manage cache invalidation.
Architecture
Application → DAX Cluster → DynamoDB Tables
DAX vs. ElastiCache
DAX
- Purpose-built for DynamoDB.
- Caching is handled at the API level.
- Works transparently with DynamoDB SDK calls.
- Best for improving DynamoDB read performance without altering app code.
ElastiCache
- General-purpose in-memory store (Redis / Memcached).
- Can cache results from any source (databases, APIs, computations).
- Requires application logic for cache reads/writes.
- More flexible, but not tied to DynamoDB.
Use DAX for direct DynamoDB acceleration.Use ElastiCache for multi-source caching or custom caching strategies.