Export Data Products to the Source System

Alation Cloud Service Applies to Alation Cloud Service instances of Alation

After you enrich a data product in Alation, you can export its definition back to the source system as a semantic model. Alation converts the data product’s definition, including its tables, metrics, relationships, and filters, into the source’s native semantic object:

  • In Snowflake, Alation creates or replaces a semantic view that you can use with native Snowflake tooling such as Cortex Analyst.

  • In Databricks, Alation creates or replaces a metric view.

This lets you maintain a single, governed source of truth in Alation and push the enriched semantics back to the source, so the source’s native tooling works from the same governed definition.

In this topic:

Prerequisites

Before you can export a data product, verify the following:

  • You can edit the data product. Typically, you must be a Data Product Admin for the data product. See Configure Access in the Data Product App.

  • The data product is listed on the Marketplace. See List on Marketplace.

  • Chat is enabled for the data product. The export option appears only after Chat is enabled, because the export reuses the same source authentication that Chat uses. See Configure Chat with Data Product.

  • The data product’s data comes from a supported source: Snowflake or Databricks. This is the case when either:

    • The data product was created from a Snowflake semantic view or a Databricks metric view, or

    • All of the data product’s tables come from a single Snowflake or Databricks data source.

  • Authentication is configured for the data product. Alation uses these credentials to create the semantic object in the source. If you haven’t configured credentials yet, Alation prompts you to set them up the first time you export. See Configure Chat Authentication.

Export a Data Product

The export action creates or replaces a semantic view (Snowflake) or metric view (Databricks) based on the current definition of the data product.

Important

Exporting overwrites any changes made directly in the source to this semantic object. Alation replaces it with the definition from the data product.

To export a data product:

  1. Open the data product. You can start the export from either:

    • The data product’s catalog page, or

    • The Manage My Data Products page.

  2. Click the three-dot menu icon to open the action menu.

  3. Select Export to Snowflake or Export to Databricks, depending on the source. The export dialog opens.

  4. In the Target Semantic View, review or enter the fully qualified name where the semantic object will be created or replaced (DATABASE.SCHEMA.SEMANTIC_VIEW_NAME for Snowflake or catalog.schema.metric_view_name for Databricks). For a data product created from a cataloged Snowflake semantic view, Alation sets this value automatically.

  5. Click Sync.

    • If authentication is not yet configured for the data product, the Enter credentials dialog appears. Provide the credentials for the data source — for example, the Username and Password, or a personal access token — optionally expand Advanced Connection Settings, and click Connect. Alation saves these credentials securely and continues with the export.

Alation notifies you when the export completes. Verify the changes in the source. If the export fails, the dialog displays the error returned by the source. Correct the issue—for example, a missing target location or invalid credentials—and try again.

What Alation Exports

When you export a data product, Alation maps the semantic structure of the definition into the source’s semantic object:

  • Tables (the source table and any joined tables)

  • Dimensions, including time dimensions and facts

  • Metrics

  • Relationships (joins between tables)

  • Filters

  • The data product description, which becomes the semantic view description (Snowflake) or the metric view comment (Databricks)

  • Verified queries based on evaluations configured for data product (Snowflake only)

The data product description is exported, as noted above. However, catalog-level enrichment stays in Alation and is not exported. This includes table and column titles, the descriptions on individual tables and columns, and sample data.

Note

Databricks metric views do not include verified queries or synonyms, so Alation does not export those when you export to Databricks.

Verify the Export

After the export completes, confirm the changes in the source. Query or open the semantic view (Snowflake) or metric view (Databricks) to verify that the additions you made in Alation, such as new metrics, appear in the exported definition.

Keep a Data Product in Sync with Snowflake

When a data product was created from a Snowflake semantic view, you can have Alation update the data product’s definition from the source automatically. When auto-sync is enabled, Alation re-syncs the data product’s definition from the source each time metadata extraction runs on the source and a new version of the data product is generated. This ensures that the data product stays up to date with any changes made in Snowflake, such as new columns or metrics added to the semantic view.

Note

Auto-sync on metadata extraction is available only for data products created from a Snowflake semantic view. It is not available or applicable for Databricks metric views.

To enable auto-sync:

  1. Open the data product settings. See Open the Data Product Settings.

  2. On the Overview tab, find the Auto-sync on metadata extraction card. The card displays the location of the source semantic view.

  3. Turn on the Auto-sync on metadata extraction toggle. Alation confirms that auto-sync is enabled.

The Auto-sync on metadata extraction toggle is turned off by default.

When auto-sync is enabled and a change in the source semantic view is detected during metadata extraction, Alation generates a new draft of the data product. You can switch between drafts on the Content tab to compare the differences before listing the updated version.