A company plans to use an Amazon DynamoDB table and wants to optimize costs. The table is mostly unused in the mornings. In the evenings, read/write traffic is unpredictable and can spike very quickly. What should a solutions architect recommend?

The correct option is to create a DynamoDB table in on-demand capacity mode. This mode is ideal for workloads with unpredictable traffic patterns and significant idle periods, as you only pay for the read and write requests your application actually consumes. This directly addresses the requirement for cost optimization with unpredictable evening spikes and morning inactivity. Creating a DynamoDB table with a global secondary index is incorrect because while indexes are crucial for query performance, they don't directly manage or optimize capacity costs for unpredictable traffic. Creating a DynamoDB table with provisioned capacity and auto scaling is incorrect because while auto scaling can adapt to traffic changes, it still incurs costs for provisioned capacity even during idle times and might not react quickly enough to very sudden, sharp spikes without over-provisioning. Creating a DynamoDB table in provisioned capacity mode and configuring it as a global table is incorrect because global tables are for multi-region replication and don't inherently optimize capacity costs for a single-region, unpredictable workload; it also uses provisioned capacity which is not cost-optimal for idle periods.