Question

A SaaS company runs an event-tracking platform on a DynamoDB table that stores hundreds of millions of records. Only the most recent 30 days of data is queried by the live application, but compliance requires that all records be retained for 7 years for occasional audit queries. The table has grown to over 4 TB and storage costs are rising sharply. Audit queries are infrequent, run offline, and can tolerate multi-second latency. The team wants to reduce ongoing DynamoDB storage costs while preserving queryability of historical data. What is the MOST cost-effective approach?

Accepted Answer

Enable DynamoDB Time to Live (TTL) on the table with a 30-day expiry and use DynamoDB Streams with a Lambda function to write expiring items to Amazon S3 in Parquet, then query the archive with Amazon Athena.

Answer

Switch the table to the DynamoDB Standard-Infrequent Access table class and keep all 7 years of data in the table so audit queries continue to use the DynamoDB API.

Answer

Enable DynamoDB TTL with a 30-day expiry and rely on the deleted items being automatically archived to S3 Glacier by DynamoDB for later retrieval.

Answer

Create an AWS Glue job that periodically scans the entire table each night, copies all records to S3, and deletes records older than 30 days from DynamoDB.