SSH tunnel

SSH tunnel is a Databricks developer experience capability, introduced 2026.

A Databricks-provided SSH tunnel that attaches VS Code, Cursor, or a plain terminal to serverless, AI Runtime, or dedicated compute, so the whole development environment runs on Databricks instead of your laptop.

• The SSH tunnel does not use port 22 - it smuggles every byte over the same HTTPS endpoint the rest of the Databricks CLI already talks to, so the thing that has to survive your corporate firewall is a WebSocket upgrade, not an open SSH port. 🔗
• IntelliJ users are told to pass --auto-start-cluster=false, because JetBrains IDEs helpfully start every configured cluster the moment you launch the IDE. 🔗
• Databricks ships an official IDE extension and an official SSH tunnel, and the docs warn that the two are not yet compatible and should not be used together. 🔗

⚠ Limitations: Shared clusters aren't supported and the Databricks IDE extension can't be used alongside it; files edited outside /Workspace, /Volumes, and /dbfs are lost on cluster restart; at most 10 SSH connections per cluster; inactive sessions might drop after 1 hour; and three or more open Jupyter notebooks can cause performance or connection issues. 🔗

Category

Developer experience

Introduced

2026🔗

Announced at

Databricks blog: Run, debug, and scale Databricks workloads from your local IDE🔗

Also known as

Databricks Remote Development, Remote Development, databricks ssh connect

Verified

2026-09-13