A team needs a large point-in-time DynamoDB dataset in S3 for analytics. The table has point-in-time recovery enabled, and the team wants to avoid consuming read capacity on the live table with a full scan. Which feature best fits?

1. A strongly consistent Scan of every item through the application API.
2. Create a new global secondary index and assume its backfill writes an S3 snapshot.
3. DynamoDB TTL with an expiration timestamp on every item.
4. DynamoDB export to S3 at a supported point within the recovery window.