Post by Rajagopal Senthil Kumar

Most data engineers know how to secure data at rest. But how do you ensure that the same table returns different results to different users without duplicating data, creating multiple views, or changing the underlying Delta table? This is where Dynamic Data Masking (DDM) in Databricks Unity Catalog becomes a powerful governance capability. One of the biggest misconceptions is that masking modifies the actual data. It doesn’t. The original Delta table remains completely unchanged. At query time, Unity Catalog evaluates: • Who is accessing the data • Which group they belong to • What permissions they have • Which masking policies apply Then Databricks dynamically returns either the original value or a masked value based on the user’s access rights. In this infographic, I explored: • Column-level masking using Unity Catalog • Row-level filtering for data segregation • Query-time policy evaluation • Real examples using customer PII data • SQL commands to enable masks on tables • End-to-end lineage from query to masked result • How Databricks protects sensitive information without data duplication The most important takeaway: Original Delta tables never change. Masking happens dynamically during query execution. This makes Dynamic Data Masking an essential capability for organizations handling PII, financial data, healthcare information, and other sensitive datasets while maintaining a single source of truth. References: Databricks Unity Catalog Row Filters & Column Masks: https://lnkd.in/gMhqyU6s Databricks SQL Functions: https://lnkd.in/gRDb4cHm #Databricks #UnityCatalog #DataGovernance #DynamicDataMasking #DataSecurity #DataPrivacy #Lakehouse #DataEngineering #BigData #Analytics #CloudArchitecture #DataArchitecture #PII #SQL