Asserting first-class help of Iceberg format in Databricks Delta Sharing

With greater than 300% year-on-year utilization progress for two consecutive years, Delta Sharing is probably the most broadly adopted open protocol for Knowledge and AI sharing. Main information suppliers together with SAP, Walmart, Atlassian, and LSEG use Delta Sharing to share information with their companions and prospects throughout clouds and platforms. Right now, we’re excited to announce that Databricks Delta Sharing has first-class help for the Apache Iceberg format.

Knowledge suppliers can now share information securely and stay from Databricks to any consumer that helps the Apache Iceberg REST Catalog API. Recipients on platforms akin to Snowflake, Trino, Flink, and Spark on all clouds can all use this functionality – including to Delta Sharing’s open ecosystem. 

Moreover, we’re launching a Personal Preview that allows information suppliers to make use of Delta Sharing to share Iceberg tables managed by catalogs exterior Databricks – together with AWS Glue, Hive Metastore, Snowflake Horizon, and extra.

Collectively, you may share any new or present tables (Delta or Iceberg, managed or overseas). This builds in direction of full open interoperability. You possibly can herald Iceberg tables from any exterior catalog, handle them via Databricks and Unity Catalog, after which share them out to any recipient—whether or not they’re on Databricks, an Iceberg consumer, or a Delta consumer. This allows you to leverage Unity Catalog as your unified information governance layer, providing you with one place for all sharing. 

On this weblog put up, we are going to clarify why open information sharing is essential. We can even dive into how Delta Sharing to Iceberg shoppers work via a hands-on demo.

Why This Issues: Open vs. Closed Sharing

Most information sharing options will not be actually sharing—they’re trapping. They’re basically closed and engineered to make sure vendor lock-in, so that you solely get to share with others who’re already inside their closed ecosystems. This limits your choices, stifles innovation, and drives large, pointless information replication.

Delta Sharing is probably the most broadly adopted open normal for safe information sharing. Utilized by category-leading information suppliers, it’s constructed to help totally different clouds and platforms. Delta Sharing operates on three core ideas:

• Share any asset.
• Share with anybody.
• Share with none friction.

Including Iceberg consumer help strengthens this dedication. It helps you to share a Delta desk whereas recipients expertise it as a local Iceberg desk. Sharing occurs over the Iceberg REST API, so recipients can join from any Iceberg-compatible platform. This lets you get the very best of each worlds: Knowledge suppliers profit from superior Delta Sharing options akin to View sharing, whereas recipients obtain native Iceberg tables through the Iceberg REST API.

Recipients get safe, stay entry to the supply information. This eliminates silos and allows you to share information brazenly with anybody.

This function is right for organizations that have to share information externally with companions and prospects utilizing Iceberg shoppers, akin to these working on Snowflake or integrating with platforms like Trino, Flink, or Spark. Corporations with a number of enterprise items working throughout a number of platforms additionally profit by unblocking seamless, bi-directional information trade in multi-cloud or hybrid environments. Industries already leveraging these patterns embrace healthcare, retail, finance, ad-tech, and extra.

Interoperability: Each Supply and Vacation spot 

As a result of we imagine in full open information entry, we do not cease at sharing information to Iceberg shoppers. We at the moment are creating the subsequent evolution: sharing overseas Iceberg tables that reside in exterior catalogs akin to AWS Glue or Snowflake Horizon. We’re excited to announce the Personal Preview of Delta Sharing help of overseas Iceberg tables.

You may ask: Why share an Iceberg desk via Delta Sharing if it resides in AWS Glue or Snowflake? Why not share immediately from inside that platform?

First, by cataloging your exterior Iceberg information in Unity Catalog, you get a unified governance layer in Unity Catalog, permitting you to get full visibility and governance throughout your information property. Moreover, utilizing Delta Sharing permits you to get the better of each worlds: You profit from Delta Sharing best-in-class sharing functionalities, whereas maintaining your information in Iceberg format. This contains for instance, the power to Delta Share Views for fine-grained entry management, which isn’t supported natively by the Iceberg IRC API.

With this Personal Preview, the Databricks Lakehouse is open in each instructions. Your Lakehouse can share information to and obtain information from the rising Iceberg ecosystem.

This twin capability offers you:

• Easy Collaboration: Work collectively irrespective of which open desk format you employ (Delta or Iceberg).
• Ruled Sharing: Unity Catalog controls entry and supplies audit logs.
• Broadest Attain: Share information as each a supplier and a recipient, breaking platform partitions.

How Does It Work?

Think about your organization, Supplier Corp, makes use of Databricks and Delta Lake to handle buyer information. You want to securely share a day by day listing of product gross sales with Companion Inc, which makes use of Snowflake and prefers the Iceberg format.

Earlier than this function: Supplier Corp must manually export the information, remodel it right into a Snowflake-readable format, add it to the accomplice’s cloud storage, and arrange a fancy synchronization job. That is sluggish, expensive, includes vital admin overhead, and dangers information changing into outdated.

With Delta Sharing to Iceberg Shoppers:

1. Supplier Corp permits Iceberg reads on gross sales information through UniForm (this will embrace managed and exterior Delta tables, views, materialized views, and streaming tables), and shares it through Delta Sharing. This supplies stay entry with no duplication or re-ingestion required.
2. Companion Inc units up a easy connection in Snowflake utilizing the supplied credentials for safe authentication through short-lived bearer tokens.
3. Companion Inc’s analysts can instantly question the shared desk utilizing normal SQL, treating it like a local Iceberg desk of their Snowflake setting.
4. The info they see is at all times stay (zero-copy), and Supplier Corp maintains full safety and governance with auditing and monitoring utilizing Unity Catalog.

This makes information sharing immediate, protected, and fully format-agnostic.