Which solution meets these requirements with the LEAST operational overhead?
Configure point-in-time recovery for the table.
Use AWS Backup for the table.
Use an AWS Lambda function to make an on-demand backup of the table every hour.
Turn on streams on the table to capture a log of all changes to the table in the last 24 hours. Store a copy of the stream in an Amazon S3 bucket.
Explanations:
Point-in-time recovery (PITR) allows you to revert a DynamoDB table to any second within the last 35 days. Since the requirement specifies reverting to any point in the last 24 hours, this is the most appropriate solution with minimal operational overhead.
AWS Backup is a fully managed backup solution, but it is designed for scheduled backups and does not support point-in-time recovery. It also involves more management overhead compared to PITR.
Using an AWS Lambda function to create hourly backups would introduce additional operational complexity and overhead to manage the Lambda function, backup schedule, and storage. PITR is a simpler, fully managed alternative.
DynamoDB Streams capture changes but do not provide the ability to directly revert a table to a previous state. It would also require custom processing to reconstruct the table, which adds significant operational overhead.
I chart that the answer is:
Configure point-in-time recovery for the table.