Connections, Deliveries, and External Storage
A connection stores settings for a recipient: a database, file storage, or another supported system. A delivery links a connection to an export and determines when and how to send the result.
Configuration
- Create a connection in the appropriate section and select a supported driver.
- Enter the address and credentials with the minimum permissions needed.
- Test the connection. This confirms reachability, not that the whole import schema is correct.
- In the export, open Getting data → Configure delivery.
- Choose a recipient, write mode, trigger conditions, and, if needed, a schedule.
- Run a test delivery and inspect the result at the recipient.
Databases
Agree on the table, update key, column types, and handling of existing rows. Review proposed commands before applying a schema change. Automatically changing another database's structure is a separate action, not a consequence of merely opening its description.
Incremental delivery
A cursor sends changes without transferring the whole catalog again. Reset it deliberately: the next delivery may send a large volume again. The recipient must handle repeated writes of the same product correctly.
Storage and costs
When results are stored in your own database or S3, you pay your provider for that storage. The platform counts its own processing and data transfer time; external storage volume is not counted as platform storage. Temporary files and artifacts that remain on the platform may count separately.
A delivery failure does not necessarily mean collection failed. Check delivery history, credentials, recipient availability, write permissions, and schema compatibility.