Databricks Delta Sharing offers ways for data collaboration, providing a solution for seamless and secure data sharing across organizations and platforms, more can be read in the previous article Accelerating Data Collaboration: Unleashing Potential with Delta Sharing and the Databricks Marketplace – part 1. In this second part, we will explore a demo of Delta Sharing on Databricks, demonstrating seamless data sharing capabilities.
The following section explains the steps to create an open share, a recipient and access to the share. It is assumed that you have a workspace with Unity Catalog and Delta Sharing enabled, and have sufficient privileges to create a delta recipient and share, as explained here. The data that was shared is the taxi trip data found in the samples catalog in nyctaxi schema and the table trips. The Delta Sharing process can be created by either using the Databricks Workspace, CLIs or APIs, in this demo SQL API was used on a Notebook.
The share was first created and the trips table was added:
In the Data Explorer the new share can be found:
Next, the creation of a recipient is done, the share with the access token will be sent:
The newly created recipient can be found below:
A Delta Sharing URL is provided when the recipient is created, it can only be used once and used to download a configuration file containing a Bearer Token and the Data URL:
With these credentials, the data recipient can access the data from different mediums other than Databricks, such as PowerBI, which has a Delta Share data source connection:
Through a Python script, by installing the library Delta Sharing and also a sample startup script can be found in the Github repository of Delta Share. The credential file was used to authenticate and the table was saved in a pandas dataframe:
This demonstration highlighted how Delta Sharing promotes easy and accessible collaboration through different mediums. With data continuing to be the main driver for insights, leveraging Delta Sharing on Databricks emerges as a valuable service for data-driven decision making.