A latency-sensitive trading platform uses Amazon DynamoDB (on-demand mode). The solutions architect must improve performance and ensure high availability while achieving minimal latency. Which option meets these requirements?

A three-node DAX cluster provides high availability and fault tolerance, which is crucial for a trading platform. DAX is an in-memory cache that significantly reduces read latency for frequently accessed data, meeting the minimal latency requirement. Configuring the application to read through DAX leverages this caching benefit. Writing directly to DynamoDB ensures immediate data consistency and durability, as DAX is primarily a read-through cache and not designed for write-through operations for critical data. Creating a two-node DAX cluster offers less fault tolerance than three nodes. Reading and writing through DAX is incorrect because DAX is optimized for reads, and writing through it introduces unnecessary overhead and potential consistency issues for writes. Reading directly from DynamoDB while writing through DAX defeats the purpose of DAX for read performance. A single-node DAX cluster lacks high availability.