A company will use Amazon DynamoDB as the primary database, but incoming data items can exceed DynamoDB item size limits (current maximum size 700 KB and growing). Which solution handles large data sizes with the most operational efficiency?

Storing large data as objects in Amazon S3 and referencing them from DynamoDB is the most operationally efficient solution. S3 is designed for cost-effective storage of large, unstructured data, and integrating it with DynamoDB allows you to leverage DynamoDB's fast, low-latency access for metadata while offloading the large data itself. This pattern is common for handling items exceeding DynamoDB's 400 KB (not 700 KB) item size limit. Filtering data with Lambda and storing in DocumentDB adds unnecessary complexity and cost for this specific problem. Splitting data into multiple DynamoDB items increases retrieval complexity and doesn't fundamentally solve the large item size issue if individual parts are still large. Compressing data with gzip might reduce size but doesn't guarantee fitting within the limit and adds processing overhead, potentially impacting performance.