Remote Fabric Jupyter & Local VS Code – Ep. 497

In this episode of the Explicit Measures podcast, Mike and Tommy dive deep into a technical topic that’s gaining traction in the Microsoft Fabric community: connecting your local VS Code installation to remote Fabric Jupyter notebooks and the Spark engine.

The Power of Local Development with Remote Compute

VS Code is an application you install locally on your machine, but with the right extensions, you can remotely connect to the Spark engine and notebooks that live inside Microsoft Fabric. This gives you the best of both worlds: the familiar, powerful VS Code development environment with the scalability of Fabric’s cloud compute.

Why This Matters

There are several compelling reasons to adopt this workflow:

Better development experience – VS Code offers superior editing, debugging, and extension support compared to the browser-based notebook interface
Cost optimization – You can develop and test locally, then only use Fabric compute when you need to run against production data
Offline development – Work on your notebooks even when you don’t have access to the Fabric service
Version control – More easily integrate your notebooks with Git and your existing DevOps workflows

The Extension Ecosystem

There’s already an extension available today that helps bridge the gap between your local VS Code and Fabric notebooks. The extension enables you to:

Connect to your Fabric workspace
Pull down notebooks for local editing
Run code against the remote Spark engine
Push changes back to the service

Cost Considerations

Mike and Tommy discuss an interesting pattern around cost management. When you’re developing notebooks, you’re consuming Fabric capacity units (CUs) every time you spin up a Spark session. By developing locally and only connecting to the remote Spark engine when necessary, you can significantly reduce your development costs.

The Local Spark Pattern

One community member has even created a Git repo that demonstrates an advanced pattern:

Mirror files from your OneLake to your local machine
Run a local Spark engine for development
Test and iterate on your notebooks offline
Publish the finished notebook back to Fabric for production use

This approach is particularly valuable for teams doing heavy notebook development who want to minimize their Fabric compute costs during the development phase.

News Roundup

Modern Visual Tooltips – Generally Available

The episode also covers the news that Modern Visual Tooltips are now generally available in Power BI. This feature has been in preview since around May 2021 – a nearly five-year preview period! The modern tooltips provide a cleaner, more informative hover experience on visuals.

DAX Performance Tip from Chris Webb

The team highlights an article from Chris Webb about a performance optimization involving CALCULATE with filters on zero. If you’re filtering a numeric column where one of the filter values is zero, there’s a potential performance consideration worth understanding. Check out Chris’s blog at Cross Join for the full details.

Watch the Full Episode

For the complete technical walkthrough and discussion, watch the full episode above. Mike and Tommy go into much more detail on the setup process, troubleshooting tips, and best practices for this workflow.

Links mentioned in this episode: