A quick-service restaurant chain runs a point-of-sale and management backend on AWS that uses a DynamoDB table in provisioned throughput mode. The table is sized for peak traffic using 100,000 read capacity units and 80,000 write capacity units. Daily traffic is predictable: very high for about 4 hours and lower the rest of the day. The company wants to lower DynamoDB costs and reduce operational overhead for IT. Which solution meets these goals most cost-effectively?

Enabling DynamoDB auto scaling is the most cost-effective solution. Auto scaling dynamically adjusts read and write capacity units (RCUs and WCUs) based on actual traffic patterns, ensuring sufficient capacity during peak hours while scaling down during off-peak times. This eliminates the need to provision for peak traffic 24/7, significantly reducing costs. It also lowers operational overhead by automating capacity management. Reducing provisioned RCUs and WCUs would lead to throttling during peak hours, impacting application performance. Switching to on-demand capacity mode could be more expensive than auto scaling for predictable workloads with distinct peak and off-peak periods, as on-demand charges per request and might not offer the same cost savings as scaling down provisioned capacity. Purchasing reserved capacity for the 4-hour daily peak would mean paying for unused capacity during the remaining 20 hours, making it less cost-effective than auto scaling.