DynamoDB Backup/PITR Operations
Understand how to restore DynamoDB tables from on-demand and point-in-time backups by provisioning new tables with appropriate settings. Learn to verify data recovery and access permissions, plan application cutover, and document restoration steps for incident management and future auditing.
Restoring DynamoDB from an on-demand backup or a point-in-time backup provisions an entirely new table resource, with its own table ARN, billing mode, and capacity settings, and IAM resource identity, rather than rewinding the original table in place. That boundary changes what "recovered" actually means during an incident, because success requires two separate things to be true: data was copied to the intended recovery point, and the new table is usable by the principals that need to read it. Treating restore as only a data recovery event misses the second half of that requirement, since a table can hold perfectly correct data and still be unreachable by the application that needs it.
That's why restore is best treated as both a provisioning event and a data recovery event, with evidence collected for each before calling anything recovered.
Choose the right recovery primitive
When bad writes happened, or data was deleted, the first decision is which recovery ...