A company stores device data in an Amazon DynamoDB table that backs a customer-facing website. The table uses provisioned read and write capacity. The company needs to compute daily performance metrics on the device data with minimal impact on the table’s provisioned capacity. Which solution meets these requirements?

An AWS Glue job with the AWS Glue DynamoDB export connector is the most suitable solution. This connector allows for an efficient, parallel export of DynamoDB data to Amazon S3 without consuming provisioned read capacity units (RCUs) from the DynamoDB table. Once the data is in S3, the Glue job can then process it to calculate performance metrics, minimizing impact on the live, customer-facing DynamoDB table. Using an Amazon Athena SQL query with the DynamoDB connector would directly query the DynamoDB table, consuming RCUs and potentially impacting the customer-facing website. An Amazon Redshift COPY command is used to load data into Redshift, not to extract from DynamoDB for general processing. An Amazon EMR job with an Apache Hive external table would also directly query DynamoDB, consuming RCUs and potentially affecting performance.