Databricks Unity Catalog

Databricks Unity Catalog data catalog

Unity Catalog registers every Databricks asset automatically, federates external catalogs, and added domains, certification, Pages and metric views in 2026.

Compare: data catalog tools

• Data Catalog: Catalog Explorer is the interface for discovering and managing data and AI assets registered in Unity Catalog (docs).

• Automated Metadata Harvesting: Native objects register automatically, and Lakehouse Federation mirrors external catalogs (MySQL, PostgreSQL, Snowflake, BigQuery, Glue, Hive) (docs).

• Search & Discovery: Objects are searchable from the workspace search bar, including search by tag keys and tag values (docs).

• AI-Powered Search: The workspace search bar supports keyword and semantic search over table names, column names, and comments, and Catalog Explorer can generate AI comments for tables (docs).

• Usage & Popularity Statistics: The Insights tab shows table and column popularity plus frequent users, queries, notebooks and dashboards (docs).

• Custom Fields & Templates: Tags provide custom metadata on securable objects, including governed tags and system tags (docs).

• Asset Certification: The `system.certification_status` tag (certified/deprecated) covers tables, views, models, dashboards, Genie Agents, apps and notebooks (docs).

• Catalog Usage Analytics: Governance Hub shows asset usage, tag coverage, classification and DQ coverage account-wide (Beta) (docs).

• Business Glossary: Pages define business concepts with owner, synonyms, domain/subdomain hierarchy and owner-accepted edits, though without a formal approval gate (Beta) (docs).

• Metadata Propagation: Tags applied at one level of the Unity Catalog object model automatically apply to all objects beneath it, though documented inheritance is scoped to ABAC policy evaluation (docs).

Databricks Unity Catalog data lineage

Unity Catalog captures table and column lineage automatically for queries that run on Databricks.

Compare: data lineage tools

• Data Lineage: Automatically tracks how data and AI assets flow and connect, from source data through to models, services, and dashboards (docs).

• Table-Level Lineage: Lineage is tracked automatically across registered tables and other securables (docs).

• Column-Level Lineage: Lineage is captured automatically down to the column level for Spark DataFrame and Databricks SQL queries, and is queryable in system.access.column_lineage (docs).

• End-to-End Lineage: Lineage spans source data through to models, services, and dashboards (docs).

• Impact Analysis: Lineage is used for impact analysis: identify downstream tables, columns, jobs, and dashboards before changing or deleting a table or column (docs).

• OpenLineage Support: External lineage is UI/API/SDK only and doesn't mention OpenLineage (docs).

Databricks Unity Catalog data stewardship

• Metadata Automation: Tag automations assign or remove governed tags by condition and keep them current (Beta) (docs).

• Data Stewardship: Not yet researched.

• Review & Approval Workflows: Access requests route to owners (GA 2026-10-28), but approval happens outside UC, with no workflow engine (docs).

• Governance Campaigns: Tag automations apply tags at scale, but there are no steward campaigns (docs).

• Critical Data Elements: Not documented (docs).

• Governance Scorecards: Governance Hub shows metrics but no questionnaire-based maturity scoring (docs).

Databricks Unity Catalog data classification

Compare: data classification tools

• Data Classification: Automatically classifies and tags sensitive data in the catalog (docs).

• Sensitive Data Discovery: Automatically classifies and tags sensitive data in the catalog (docs).

• Data Policy: Access managed through privileges, attribute-based policies, row and column filters, and workspace bindings (docs).

• Data Masking: Row and column filters restrict which rows a principal sees and how column values are returned (docs).

• Native Policy Enforcement: Unity Catalog enforces policy within Databricks, the platform it is part of. Under this row's definition that is a native governance layer governing its own platform, not cross-platform push-down into systems the tool does not operate (docs).

Databricks Unity Catalog AI features

Genie answers questions about your data, and Unity Catalog governs AI models, agents and gateways.

• AI Chat Assistant: Genie One answers natural-language questions, filters by domain and uses Pages and certification as context (docs).

• AI Documentation Generation: AI-generated table and column comments in Catalog Explorer, with a language selector since 2026-04-22 (docs).

• Agentic Metadata Enrichment: Data Classification uses "an agentic AI system" to classify and auto-tag sensitive columns (GA 2026-04-20) (docs).

• AI Features: Not yet researched.

Databricks Unity Catalog data compliance

• Data Compliance: Maintains a complete record of all data access and system activity using the audit log system table (docs).

• Audit Trail: Audit log system table maintains a complete record of all data access and system activity (docs).

• Data Activity Monitoring: Audit and query-history system tables record access, but no anomaly analysis is built in (docs).

• Compliance Reporting: Compliance security profiles harden workspaces, but no compliance reports are documented (docs).

• Regulatory Framework Mapping: Same source.

• ROPA: Not documented.

• Data Retention Policies: Only system-table retention (Beta) and a dropped-table recovery period; no policy-based deletion of regulated data.

• Data Subject Rights: Not documented as a feature.

Databricks Unity Catalog collaboration

• Alert Notifications: Access requests go to Slack/Teams channels, and DQ anomaly alerts are in Beta (docs).

• Collaboration: Not yet researched.

• Discussions & Annotations: Comments are descriptions, not threads.

• Ticketing Integrations: No Jira or ServiceNow destination; generic webhooks only.

Databricks Unity Catalog data marketplace

• Self-Service Access Requests: Request for access went GA on 2026-10-28. Same source as Alert Notifications.

• Data Mesh Support: Domains (GA 2026-10-08) are a business-aligned layer, and catalog ownership is delegated per team (docs).

• Data Marketplace: Not yet researched.

• Data Products: The Discover page curates assets by domain but doesn't use "data products" (docs).

• Subscription Management: Not documented.

Databricks Unity Catalog data quality

• Data Quality: Built-in profiling and alerts track data health and catch anomalies before they reach downstream consumers (docs).

• SQL Table Tests: Pipeline expectations, including on standalone materialized views (2026-06-23) (docs).

• Anomaly Detection: Alerts catch anomalies before they reach downstream consumers (docs).

• Data Profiling: Built-in profiling is part of data quality monitoring (docs).

• Data Health Scores: Per-table health indicators (Preview) and "% healthy tables" are binary, not composite scores (docs).

• Data Contracts: Not in UC; only DQX (Labs) imports ODCS.

Databricks Unity Catalog AI governance

• AI Governance: Governs AI assets and AI traffic using Unity Catalog together with AI Gateway — model permissions, consumption limits, content policies, and inference logging (docs).

• AI Asset Inventory: AI assets are registered in Unity Catalog alongside data assets and managed through Catalog Explorer; models are securables with the same privilege model as tables and volumes (docs).

• AI Lineage: Lineage tracks how data and AI assets connect, from source data through to models, services, and dashboards (docs).

• AI Agent Access Governance: AI Gateway controls agent tool access through governed securables and gates MCP service access using the same privileges applied to tables and volumes; service policies govern the content of requests and responses to AI services, with tool filtering for MCP servers (docs).

• AI Regulatory Compliance: Unity Catalog registers and governs models through the MLflow model registry, but ships no EU AI Act, NIST AI RMF or ISO 42001 risk-classification or evidence workflows (docs).

• AI Model Health Score: Model versions carry lifecycle aliases and tags, but there is no composite trust or health score per model (docs).

Databricks Unity Catalog workflows

• REST / GraphQL API: Unity Catalog REST API (docs).

• MCP Server: MCP services are governed securables — access is granted and revoked with Unity Catalog privileges, and tool filtering is applied to MCP servers (docs).

• Python SDK: databricks-sdk on pip (docs).

• Workflows: Not yet researched.

• Webhooks: Access-request webhooks exist, but no general catalog-event webhooks.

• dbt Integration: No dbt manifest ingestion; lineage from dbt runs is captured as SQL lineage.

• Browser Extension: Not documented.

Databricks Unity Catalog data context layer

• Semantic Models: Metric views define governed KPIs, dimensions and relationships (docs).

• Data Context Layer: Not yet researched.

• Knowledge Graph: Not documented.