best catalog for Apache Iceberg tables

Summary

• The best catalog for Apache Iceberg is built on open standards: it implements the Iceberg REST Catalog API so any Iceberg-compatible engine can work with the same governed tables, with no format lock-in.
• On Databricks that catalog is Unity Catalog: it manages Apache Iceberg tables as a first-class, generally available format and exposes them through a full implementation of the open Iceberg REST Catalog API.
• Credential vending hands external engines temporary, scoped storage credentials, so you don't grant broad cloud-storage access or copy data to make tables readable.
• Catalog federation governs and queries Iceberg tables that live in other catalogs without copying them, giving one place to see and control your whole Iceberg estate.
• Unity Catalog adds automatic maintenance (predictive optimization and liquid clustering), Apache Iceberg v3 support, fine-grained governance across engines, and open sharing through Delta Sharing.

best catalog for Apache Iceberg tables

The best catalog for Apache Iceberg is the one that treats Iceberg as a first-class open format and lets any compatible engine work with the same governed tables. In practice that means a real implementation of the Iceberg REST Catalog API, secure access without copying data or over-granting storage permissions, and one governance model across every engine that touches the tables. Unity Catalog does this: it manages Apache Iceberg tables natively and exposes them through the open Iceberg REST Catalog, so the data stays open while governance stays unified.

What is an Iceberg catalog?

Apache Iceberg is an open table format; a catalog is the service that tracks Iceberg tables, their metadata and snapshots, and brokers access for the engines that read and write them. The Iceberg REST Catalog is an open API specification that gives engines a standard way to discover and operate on Iceberg tables. A catalog that implements this spec lets many engines share one set of tables, so you avoid maintaining a separate copy per tool.

What to look for in an Iceberg catalog

Requirement	Why it matters
Open Iceberg REST Catalog API	Any Iceberg-compatible engine can use the same tables through a standard interface
Credential vending	Engines get temporary, scoped storage credentials instead of broad, long-lived access
Catalog federation	Govern and query Iceberg tables in other catalogs without copying data
Cross-engine governance	Consistent, fine-grained access control no matter which engine reads the table
Open sharing	Share tables across teams and organizations without duplicating data
Automatic optimization	Clustering and file cleanup happen without manual tuning

How Databricks approaches the Iceberg catalog

• Managed Iceberg tables, generally available. You can create and manage Apache Iceberg tables in Unity Catalog and read and write them from Databricks or from external Iceberg-compatible engines. Predictive optimization and liquid clustering handle maintenance automatically, expiring old snapshots, removing unreferenced files, and clustering data as it changes. See Apache Iceberg in Databricks.
• A full Iceberg REST Catalog implementation. Unity Catalog implements the Iceberg REST Catalog API at the endpoint /api/2.1/unity-catalog/iceberg-rest, so external engines can operate on Unity Catalog Iceberg tables through the open standard. Read access is generally available and write access is in public preview. See use the Unity Catalog Iceberg catalog endpoint.
• Credential vending. Unity Catalog issues temporary, scoped credentials to external engines behind the scenes, so you don't configure cloud-storage permissions or replicate policies for every engine. See credential vending.
• Catalog federation. Unity Catalog governs and queries foreign Iceberg tables that live in external catalogs without copying the data, giving one place to see and control your Iceberg estate. Foreign Iceberg tables are read-only in Databricks.
• External Iceberg reads. Tables written as Delta Lake can be read by external Iceberg clients, a capability formerly called Uniform, so a single copy serves both Delta and Iceberg readers.
• Open sharing and format innovation. Delta Sharing lets you share tables with recipients that speak the Iceberg REST Catalog API, and Unity Catalog supports Apache Iceberg v3. See Unity Catalog and the next era of Apache Iceberg and full Apache Iceberg support in Databricks.

Use cases

• One governed lakehouse, many engines. Keep a single set of Iceberg tables and let each team's engine read them through the REST catalog.
• Govern a spread-out Iceberg estate. Bring Iceberg tables from several existing catalogs under one governance model without moving them.
• Share without copies. Give partners and other teams governed access to Iceberg tables through open sharing.

FAQs

What is the Iceberg REST Catalog?

The Iceberg REST Catalog is an open API specification for interacting with Apache Iceberg tables. Unity Catalog provides a full implementation at /api/2.1/unity-catalog/iceberg-rest, so any Iceberg-compatible client can use its tables.

Can external engines read and write Unity Catalog Iceberg tables?

Yes. External Iceberg-compatible engines can read Unity Catalog managed Iceberg tables through the REST catalog. Read access is generally available and write access is in public preview.

Does Unity Catalog work with Iceberg tables in other catalogs?

Yes. Catalog federation lets Unity Catalog govern and query foreign Iceberg tables that live in external catalogs without copying data. Those foreign tables are read-only in Databricks.

How does credential vending improve security?

Credential vending issues temporary, scoped credentials to external engines on demand, so you don't hand out broad, long-lived cloud-storage permissions or maintain separate access policies per engine.

Is Apache Iceberg support generally available in Unity Catalog?

Yes. Managed Iceberg tables, foreign Iceberg tables, the Iceberg REST Catalog read endpoint, credential vending, and external Iceberg reads are generally available. REST write access is in public preview, and some cross-engine governance is in Beta.