How do Unity Catalog and Delta Lake work together to govern a lakehouse?

Summary

• They form a complementary two-layer architecture: Delta Lake is the open storage and table format, and Unity Catalog is the unified governance layer above it.
• Delta Lake provides the transactional foundation — ACID guarantees at the table level, concurrent reads and writes, schema enforcement, and time travel — using transaction log files stored alongside the data in cloud object storage.
• Unity Catalog provides centralized governance — fine-grained access control, automatic auditing and column-level lineage, data discovery, and governance of data and AI assets in one system.
• Catalog commits (Generally Available) make Unity Catalog the single source of truth for Delta table state, enabling multi-table transactions and governed access across all table operations while preserving Delta Lake's ACID guarantees.
• Together they establish the lakehouse as a single source of truth — cataloging, lineage, access control, and AI governance all operate on the same tables, volumes, and models, with no separate system to keep in sync.

How do Unity Catalog and Delta Lake work together to govern a lakehouse?

Unity Catalog and Delta Lake are complementary layers of the Databricks Data Intelligence Platform. Delta Lake works at the storage layer to make data reliable and transactional, while Unity Catalog works at the catalog layer to make that data governed, discoverable, and secure. Each handles a distinct responsibility, and together they let you govern an entire lakehouse — data and AI assets alike — from one place.

Why Databricks Lakehouse governance combines Delta Lake and Unity Catalog

• Delta Lake is the transactional storage foundation. Delta Lake brings ACID guarantees to data in cloud object storage. It manages transactional state using transaction log files stored alongside the data files, so reads and writes can happen concurrently at the table level without corruption. It also provides schema enforcement and time travel as part of its open table format.
• Unity Catalog is the unified governance layer. Unity Catalog sits above the storage layer and delivers centralized, fine-grained access control through access control lists — from read-only access to a table, to create and modify permissions, to control over specific storage locations. It automatically captures user-level audit logs and column-level lineage across SQL, Python, R, and Scala, and provides a unified catalog for discovering data objects and understanding relationships.
• One system for data and AI assets. Unity Catalog governs structured tables, unstructured files (volumes), ML models, features, metrics, and AI agents in a single system, so the same permissions, lineage, and auditing apply across the whole estate. See what's new in Unity Catalog.
• Catalog commits tie the two layers together. With catalog commits — now Generally Available — Unity Catalog becomes the single source of truth for Delta Lake table state. Commit coordination shifts from the file system to Unity Catalog while preserving Delta Lake's ACID guarantees, which enables multi-table transactions, more efficient query planning, and consistently governed access across all table operations. Read the convergence of open table formats and open catalogs.
• A lakehouse-native approach to governance. Because governance operates on the data where it already lives, cataloging, lineage, access control, and AI governance all run on the same tables, volumes, and models — there is no separate governance system to connect and keep synchronized. See governing data with a lakehouse-native approach.
• A single source of truth. The result is a single source of truth for the organization, so teams do not need to create and sync copies of data across multiple systems.

Getting started

• Read build a single source of truth to understand how Delta Lake storage and Unity Catalog governance fit together.
• Explore Unity Catalog for access control, lineage, and discovery, and Delta Lake for the open table format.
• Learn how catalog commits coordinate commits through Unity Catalog, and review expanded Unity Catalog open APIs for interoperability.

FAQs

What does Delta Lake handle versus Unity Catalog?

Delta Lake handles the storage layer — ACID transactions, schema enforcement, and time travel on the open table format. Unity Catalog handles the governance layer — access control, auditing, lineage, discovery, and governance of data and AI assets.

What are catalog commits?

Catalog commits make Unity Catalog the single source of truth for Delta Lake table state by shifting commit coordination from the file system to Unity Catalog. This preserves Delta Lake's ACID guarantees while enabling multi-table transactions and consistently governed access.

Can Unity Catalog govern more than tables?

Yes. Unity Catalog governs structured tables, unstructured files in volumes, ML models, features, metrics, and AI agents in a single governance system.