Delta Sharing fundamentals

Introduction to Databricks Lakehouse

Gang Wang

Senior Data Scientist

The data sharing problem

$$

  • Traditional sharing: export, email, upload
  • Data goes stale immediately
  • Multiple copies across organizations
  • No control once the file leaves your hands

$$

recraft: half: A person struggling to carry multiple heavy boxes of files through a door, representing the burden of traditional data sharing through file exports

Introduction to Databricks Lakehouse

What is Delta Sharing?

$$

architecture: Delta Sharing Protocol

$$

  • No copies - recipients query live data through the Delta Sharing protocol
  • Revoke instantly - remove access with a single command
Introduction to Databricks Lakehouse

Creating a share

$$

-- Create a share
CREATE SHARE partner_data;

-- Add a gold table to the share
ALTER SHARE partner_data
ADD TABLE production.sales
  .gold_daily_revenue;

$$

  • A share is a named collection of tables
  • Add tables from any catalog in Unity Catalog
  • Typically share gold layer data - clean, aggregated, business-ready
Introduction to Databricks Lakehouse

Adding recipients

$$

-- Create a recipient
CREATE RECIPIENT acme_corp;

-- Grant the recipient access
GRANT SELECT ON SHARE partner_data
TO RECIPIENT acme_corp;

$$

  • A recipient represents an organization or person
  • They receive an activation link to set up access
  • You control which shares each recipient can see
Introduction to Databricks Lakehouse

The library card analogy

$$

  • Instead of photocopying a book (exporting data), you issue a library card (a share)
  • The recipient reads the original in real time
  • You can revoke the card at any time

$$

recraft: half: A person handing a library card to another person in front of a grand library building, representing granting live data access instead of copying data

Introduction to Databricks Lakehouse

Summary

$$

  • Delta Sharing provides live, read-only access to data - no copies
  • Create a share, add gold-layer tables, grant access to recipients
  • Recipients query data in real time through the Delta Sharing protocol
  • Revoke access instantly when the relationship ends
Introduction to Databricks Lakehouse

Let's practice!

Introduction to Databricks Lakehouse

Preparing Video For Download...