A company uses Amazon DynamoDB to store sensor data from IoT devices. The table has a partition key of DeviceID (String) and a sort key of Timestamp (Number). The table is configured with provisioned capacity of 1000 read capacity units (RCUs) and 500 write capacity units (WCUs). Auto Scaling is enabled with target utilization of 70% and is working correctly. Recently, a new batch job was introduced that performs Scan operations on the entire table every hour. The Scan consumes many read capacity units and causes throttling of the sensor write requests. The team wants to minimize the impact on the write-heavy sensor ingestion. Which approach is BEST?
On-demand capacity mode automatically scales to handle any traffic burst, eliminating throttling for both reads and writes. It is the best choice to prevent scan-induced throttling of write requests.
Why this answer
Switching to on-demand capacity mode eliminates throttling by automatically scaling read/write capacity to meet traffic demands. Since the scan operation runs hourly and consumes significant RCUs, on-demand mode will handle the burst without throttling write requests. Option C is incorrect because DAX caches individual items, not entire scan results; hourly scans would cause cache misses, providing no benefit.
Option A is incorrect because increasing write capacity does not address read throttling. Option B is incorrect because off-peak scheduling may not be feasible and does not guarantee elimination of throttling.