A company deployed an application that uses Amazon DynamoDB. The operations team measured expected load and provisioned the table's RCUs and WCUs to match the anticipated peak. The workload has a weekly 4-hour peak that is double the average traffic; for the remainder of the week, traffic is near the average. The access pattern is write-heavy (many more writes than reads). You must recommend a cost-minimizing approach that meets these requirements. Which solution should you choose?

The correct solution minimizes cost while accommodating the workload. Purchasing reserved capacity for the average load provides a significant discount for the majority of the week. AWS Application Auto Scaling then handles the predictable weekly 4-hour peak by temporarily increasing provisioned capacity, incurring standard provisioned rates only for that short duration. Configuring on-demand capacity mode would be more expensive because it charges per request, which is generally higher than provisioned capacity, especially for predictable workloads. Deploying DAX is primarily for read-heavy workloads and would not address the write-heavy nature of this application or the cost-effectiveness for the write-heavy peak. Reducing provisioned read capacity with DAX would also likely lead to throttling for reads, and DAX doesn't help with write capacity. Switching to on-demand with DAX still incurs the higher on-demand costs and doesn't optimize for the write-heavy peak.