Collaboration and version control

Introduction to Databricks Lakehouse

Gang Wang

Senior Data Scientist

Sharing notebooks

Grant workspace permissions to individuals or groups
Three levels: Can Read, Can Run, Can Edit
Share via the workspace UI or a direct URL

Permission	View	Execute	Modify
Can Read	Yes	No	No
Can Run	Yes	Yes	No
Can Edit	Yes	Yes	Yes

Collaboration features

Comments on specific cells for code review
Version history shows who changed what and when
But version history is linear - no branching or merging

collaboration: Two laptops connected by digital lines with comment bubbles and version history icons

Enter Databricks Repos

Databricks Repos workflow

Git integration directly in the workspace
Connect to GitHub, GitLab, Bitbucket, Azure DevOps
Clone, commit, push, pull, branch, merge
Full CI/CD workflows from within Databricks

The CI/CD workflow

flowchart: Feature Branch, Pull Request, Main Branch, Production

Notebook versioning vs. Repos

comparison: Versioning, Linear history, No branches, No external collab, Zero setup | Repos, Full Git, Branches and PRs, External teams, Provider setup

Summary

Share notebooks with Can Read, Can Run, or Can Edit permissions
Built-in version history is linear - no branches
Databricks Repos brings full Git integration to the workspace
Repos enables CI/CD workflows with branches, PRs, and automated deployment

Let's practice!

Introduction to Databricks Lakehouse