Unity Catalog OSS vs managed Unity Catalog, what's the difference

Summary

• Unity Catalog open source (UC OSS) is a self-hosted, Apache 2.0-licensed catalog providing the core catalog APIs and credential vending, deployable via Docker, Kubernetes, or bare server.
• Managed Unity Catalog in Databricks includes the open-source core and adds automated lineage, audit system tables, data quality monitoring, AI governance, OpenSharing, and a broader set of governed asset types.
• Both editions expose compatible open APIs: the same REST spec, Apache Hive's metastore API, and Apache Iceberg's REST catalog API, letting external engines access governed data.
• UC OSS is governed by the LF AI & Data Foundation and hosted at github.com/unitycatalog/unitycatalog; features including Views, Delta Sharing, Models, and enhanced access control are in development for the OSS project.
• Teams using Databricks as their primary platform get managed Unity Catalog automatically; teams building multi-engine data environments can adopt UC OSS as a shared governance layer.

Unity Catalog OSS vs managed Unity Catalog, what's the difference

Unity Catalog comes in two forms: an open-source implementation (UC OSS) and the managed service built into Databricks. Both share the same core catalog APIs and the same three-level namespace model. UC OSS provides the foundational catalog and governance APIs as a self-hosted service. Managed Unity Catalog in Databricks adds automated lineage, audit logging, data quality monitoring, AI governance, and deep integration with Databricks compute and storage. The open-source project is actively building toward broader feature parity.

What is Unity Catalog OSS?

Unity Catalog OSS is an open-source data and AI catalog released under the Apache 2.0 license, hosted by the LF AI & Data Foundation at github.com/unitycatalog/unitycatalog. It is deployable via Docker, Kubernetes, or tarball on any server. UC OSS provides the core governance APIs: a REST API built on an OpenAPI specification, compatibility with Apache Hive's metastore API, and compatibility with Apache Iceberg's REST catalog API. In its current open-source release, it governs tables, volumes (unstructured data), and functions. Views, Delta Sharing (OpenSharing), Models, Remote Functions, and enhanced access control APIs are in active development for the OSS project.

Unity Catalog OSS vs managed Unity Catalog: how they compare

Capability	Unity Catalog OSS	Managed Unity Catalog (Databricks)
Deployment	Self-hosted: Docker, Kubernetes, bare server	Fully managed, built into Databricks workspaces
License	Apache 2.0, governed by LF AI & Data Foundation	Databricks subscription service
Asset types	Tables, volumes, functions	Tables, volumes, functions, views, ML models, serving endpoints, MCP services
Core catalog APIs	OpenAPI REST spec; compatible with Hive metastore API and Iceberg REST catalog API	Same open APIs plus Databricks extensions
Access control	Credential vending; basic RBAC	Full RBAC with row/column filters, attribute-based policies, and workspace bindings
Automatic lineage	Not in current OSS release	Automatic lineage tracking across all Databricks workloads
Audit logging	Not in current OSS release	Complete audit log in system.access system tables
Data quality monitoring	Not in current OSS release	Built-in profiling and anomaly detection
AI asset governance	Not in current OSS release	ML models, serving endpoints, functions, and Unity Gateway traffic
Delta Sharing / OpenSharing	In development for OSS	Supported via OpenSharing
Automatic enablement	Manual deployment required	Automatically enabled for Databricks workspaces after Nov 8, 2023

How Databricks delivers managed Unity Catalog

Managed Unity Catalog is built into the Databricks platform and automatically enabled for workspaces created after November 8, 2023. It uses the same three-level namespace (catalog.schema.object) as UC OSS and enforces access control through SQL-standard GRANT and REVOKE statements. Additional fine-grained controls include row-level security, column masks, and attribute-based policies for users, service principals, and groups.
Lineage is tracked automatically for all Databricks workloads. Notebooks, jobs, Lakeflow Pipelines, and model training runs all produce lineage events that Unity Catalog captures without any instrumentation. The system.access system table provides a complete, queryable audit log of every access event.
AI governance is a first-class capability in managed Unity Catalog. ML models, model serving endpoints, and AI functions are governed catalog objects with the same access control and lineage as data tables. Unity Gateway, Databricks' AI traffic governance layer, integrates with Unity Catalog for policy enforcement on model traffic.
For external engines, managed Unity Catalog exposes the same open APIs as UC OSS. Tables governed in Databricks are accessible from DuckDB, Apache Spark, and other Iceberg-compatible tools through the standard Iceberg REST catalog or Hive metastore protocols.

FAQs

Can I use UC OSS independently of Databricks?

Yes. UC OSS is a self-contained server you deploy on your own infrastructure. It provides the core catalog APIs without requiring Databricks. Teams use it to provide a common catalog and credential vending layer for multi-engine environments, including DuckDB, Apache Spark, and other Iceberg-compatible tools.

Do UC OSS and managed Unity Catalog share the same namespace model?

Both use the same three-level namespace (catalog.schema.object) and expose compatible REST APIs. A catalog created in UC OSS follows the same structural conventions as one in managed Unity Catalog, and external engines interact with both through the same Iceberg and Hive-compatible protocols.

Which features are not yet in UC OSS?

As of the current open-source release, Views, Delta Sharing (OpenSharing), Models, Remote Functions, and enhanced access control APIs are still in development for UC OSS. These capabilities are available today in managed Unity Catalog within Databricks. The UC OSS GitHub repository tracks current roadmap status.

Is managed Unity Catalog available outside Databricks?

Managed Unity Catalog is part of the Databricks platform and runs as a managed cloud service. UC OSS is the path for running the catalog server independently on your own infrastructure. Both expose compatible open APIs.