How Delta Sharing Supports ABAC Sharing for Providers and Recipients
Summary
Delta Sharing now supports Attribute-Based Access Control (ABAC) for secure table sharing, allowing data providers to share assets without duplicating tables or policy logic. Recipients can apply their own ABAC policies on shared tables, provided they also apply the necessary governed tags on their side. These recipient-side policies are enforced at query time, enabling local governance and independent business logic. All actions, including policy creation, sharing, and access requests, are logged in Unity Catalog, ensuring traceability for regulatory compliance and internal audits. This capability has already been adopted by customers like Yanolja, a global travel platform, to enhance partner engagement and ensure consistent data usability across regions.
Key takeaway
For Data Engineers or MLOps Engineers managing data sharing, implementing ABAC Table Sharing via Delta Sharing allows you to distribute data securely while empowering recipients to apply their own governance. This approach reduces friction and ensures compliance without compromising control, streamlining data access for diverse consumers across multiple workspaces.
Key insights
ABAC-enabled Delta Sharing allows secure, governed data exchange with independent recipient-side policy enforcement.
Principles
- Decouple data sharing from policy duplication.
- Recipient-side tags enable local policy enforcement.
Method
Data providers build shares via Delta Sharing. Recipients apply their own ABAC policies and corresponding governed tags. Policies are enforced at query time, with all actions logged in Unity Catalog.
In practice
- Share data across regions with dynamic policies.
- Ensure compliance with local governance requirements.
Topics
- ABAC
- Delta Sharing
- Data Governance
- Unity Catalog
- Policy Enforcement
Best for: Data Engineer, MLOps Engineer, AI Security Engineer
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by Databricks.