Skip to main content
This guide walks you through creating a notebook, selecting a kernel, running cells against an approved connection, and optionally attaching the notebook to a pipeline for repeatable execution.

​
Prerequisites

  • Enterprise notebook entitlement enabled for your workspace
  • A connection (warehouse or Spark) your role may use
  • Browser permissions allowing WebSocket connections to the notebook service (corporate proxies sometimes block these)

​
Setup steps

1

Create a notebook

Open Notebooks → New notebook. Name it after the investigation (q3_churn_slice_explore) so teammates recognize the intent.
2

Pick a kernel

Choose SQL, Python, or Scala depending on connector support. Match the dialect to your warehouse when using SQL kernels.
3

Attach a connection

Select the connection from the notebook sidebar. Test connectivity with a trivial query (SELECT 1 or SELECT current_timestamp()).
4

Run cells top to bottom

Execute imports and parameters first, then analysis cells. Restart kernel if package installs or credential changes occur mid-session.
5

Save and share

Save revisions; share read-only links with collaborators who have workspace access.

​
Notebook node on the pipeline canvas

Some workspaces expose a Notebook node type that executes a saved notebook as part of a pipeline run:
  1. Author and test the notebook interactively until outputs are stable.
  2. Parameterize file paths, dates, and environment names—avoid hard-coded prod literals.
  3. Drag the Notebook node onto the canvas, select the saved artifact, and map parameters from upstream nodes or pipeline variables.
  4. Run the pipeline in dev with representative partitions before promoting.
Long-running notebook cells can block orchestration SLAs. Keep production notebook nodes focused; move heavy exploration back to interactive mode.

Spark integration

Scale notebooks on Spark clusters.

Compute

Configure execution environments for notebooks.