Forum Discussion
Microsoft Purview Data Map Approach to scan
I plan to scan Purview data assets owner by owner rather than scanning entire databases in one go because this approach aligns with data governance and RBAC (Role-Based Access Control) principles. By segmenting scans by asset ownership, we ensure that only the designated data asset owners have the ability to edit or update metadata for their respective assets in Purview.
This prevents broad, unrestricted access and maintains accountability, as each owner manages the metadata for the tables and datasets they are responsible for.
Scanning everything at once would make it harder to enforce these permissions and could lead to unnecessary exposure of metadata management rights. This owner-based scanning strategy keeps governance tight, supports compliance, and ensures that metadata stewardship remains with the right people.
This approach also aligns with Microsoft Purview best practices and the RBAC model:
- Microsoft recommends scoping scans to specific collections or assets rather than ingesting everything at once, allowing different teams or owners to manage their own domains securely and efficiently.
- Purview supports metadata curation via roles such as Data Owner and Data Curator, ensuring that only users assigned as owners; those with write or owner permissions on specific assets; can edit metadata like descriptions, contacts, or column details.
- The system adheres to the principle of least privilege, where users with Owner/Write permissions can manage metadata for their assets, while broader curation roles apply only where explicitly granted.
Therefore, scanning owner by owner not only enforces governance boundaries but also ensures each data asset owner retains exclusive editing rights over their metadata; supporting accountability, security, and compliance. After scanning by ownership, we can aggregate those assets into a logical data product representing the full database without breaking governance boundaries.
Is this considered best practice for managing metadata in Microsoft Purview, and does it confirm that my approach is correct?