A company collects data from many participants using wearable devices and stores it in a DynamoDB table. The workload is constant and predictable. The company wants to remain at or below its forecasted DynamoDB budget. Which solution is MOST cost-effective?

The correct answer is to use provisioned mode and specify RCUs and WCUs. Since the workload is constant and predictable, provisioned mode is the most cost-effective option. You pay for the capacity you provision, which aligns with a predictable budget. Using provisioned mode with DynamoDB Standard-IA is incorrect because Standard-IA is for data that is accessed less frequently and can tolerate slightly higher latency, which isn't specified as a requirement here. Reserving capacity is a separate cost optimization for provisioned mode, but the core answer is using provisioned mode itself. Using on-demand mode is incorrect because it's designed for unpredictable workloads and charges per request, which is generally more expensive than provisioned mode for stable workloads. Setting RCUs/WCUs in on-demand mode is also incorrect; on-demand scales automatically, you don't specify capacity units directly.