A loyalty platform needs transactional writes with strong consistency, and its catalog reads reach millions per second with microsecond responses for hot items. Which architecture fits?
Show answer and explanation
Answer: D
DynamoDB handles the write path, and strongly consistent reads and writes are available when requested. DAX sits in front of the table as a managed, in-memory cache. DAX serves cached item and query responses in microseconds and scales to very high read volumes. Read replicas and ElastiCache can add read capacity, but they require separate cache management and replication design rather than a cache built for DynamoDB.
Why the other options are wrong
- A. RDS read replicas offload reads, but replica lag means reads are not always current, and per-query latency is far above microseconds at this scale.
- B. Aurora replicas scale reads well, though each replica still runs database queries at millisecond latency, which does not meet the microsecond target.
- C. A cache-aside layer on ElastiCache requires custom code for invalidation and does not give a managed, DynamoDB-native read path.