Databricks Dataset Materialization

AI/BI Dashboards

In this article, we’ll explore how dataset materialization works in AI/BI Dashboards and how you can enable it.

By default, each dataset is evaluated live whenever a user interacts with a dashboard visual. When a user clicks or interacts with a visual, the underlying SQL query is triggered in the backend, the results are retrieved, and the visualization is then updated.

To improve dashboard responsiveness and reduce repeated query execution, dataset materialization can be enabled.

You can enable dataset materialization for an AI/BI Dashboard from the Previews section in Databricks.

In this section, let’s understand how dataset materialization works through a practical use case. When selecting data from Unity Catalog, AI/BI Dashboards provide two options:

1. Select Table
   Selecting a table creates local metric views, which are evaluated when the dashboard runs.
2. Select SQL
   Selecting SQL allows you to define a query whose results can be pre-computed and stored. This enables dataset materialization, allowing the dashboard to retrieve the stored results…