Question: How do you dump multiple DynamoDB tables?

Answer

Export each DynamoDB table to S3 separately, then track all export jobs to completion. DynamoDB does not provide a single database-wide dump operation covering multiple tables.

For large tables, use ExportTableToPointInTime. Enable PITR on each table first, choose an export time within its recovery window and send its output to a distinct S3 prefix. Save each returned export ARN and inspect DescribeExport until the job completes or reports an error. A common export timestamp helps align copies, but it does not create a cross-table transactional snapshot.

Exports produce DynamoDB JSON or Amazon Ion and do not consume table read capacity. Record table definitions, keys, indexes and operational settings separately; data files alone do not recreate the whole environment. Native exports can also be incremental or filtered when you need changes or a subset.

For small tables, a script can paginate Scan and save the results locally. Handle every LastEvaluatedKey, limit concurrency and account for read charges. A scan is not a point-in-time snapshot, even with strong reads.

Validate sample records and manifests before treating the dump as usable. See export to S3 and JSON exports for formats and restore planning.

Source: AWS export behavior.