An ecommerce company stores terabytes of customer data in AWS, including personally identifiable information (PII). Three applications will use the data; only one application needs the PII. The PII must be removed before the other two applications process the data. Which solution meets these requirements with the LEAST operational overhead?

S3 Object Lambda allows you to process and transform data retrieved from S3 buckets as it is being accessed, without modifying the original objects. This means the PII can be removed on-the-fly for the two applications that don't need it, while the original data remains intact for the application that does. This approach minimizes operational overhead because you don't need to manage separate datasets or complex proxy layers. Storing data in DynamoDB is less suitable for terabytes of unstructured customer data. Creating separate S3 buckets or DynamoDB tables for each application would involve duplicating data and managing multiple datasets, increasing operational complexity and storage costs. A proxy application layer would also add operational overhead for development and maintenance.