The Dataprep service provides a web-interface for users to define the transformation logic and schedule job execution. Dataprep anonymously and securely stores this logic in the form of metadata within Google Cloud SQL. The Dataprep service itself does not transform nor store any of the customer’s physical data.
The Dataprep service translates user-generated metadata describing data transformation logic into a Dataflow job or BigQuery SQL statements which reads, transforms and writes the customer data exclusively within the user’s Google Cloud project. During job execution processed by Dataflow or BigQuery, data does not leave the customer’s Google Cloud project.
The Dataprep service only exists within the Google Cloud ecosystem, and can only be launched from the Google Cloud console after authorization through Google Cloud Identity and Identity Access Management (IAM) roles.