Connect Power BI, Tableau, Qlik

The BI tool is the last mile: it’s where the classifications, ownership, and PII tags Kirimana wrote become something an analyst sees on the canvas. Kirimana materialises governed gold tables through your dbt build into a Databricks SQL warehouse, registers them in Unity Catalog, and your BI tool reads them from there. This page covers pointing each of the three most-asked tools at that warehouse, and what the contract metadata gives the BI layer once it’s connected.

All three tools connect the same way at the core: a native Databricks connector against a SQL warehouse, authenticating to Unity Catalog, reading the gold catalog and schema Kirimana governs. Have three things ready before you start:

• Your workspace URL (e.g. https://adb-<workspace-id>.<region>.azuredatabricks.net).
• The SQL warehouse HTTP path — under SQL Warehouses → your warehouse → Connection details. Any running or serverless warehouse works; the default Starter Warehouse is fine to begin with.
• Credentials — a personal access token, or OAuth / your organisation’s SSO if the tool supports it against Databricks.

Once connected, browse to the catalog and schema that hold your Kirimana-governed gold tables and select the models your dbt build materialised.

Power BI

Power BI ships a native Databricks connector.

1. Get Data → Databricks (or Azure Databricks).
2. Fill in the Server Hostname (your workspace host) and the HTTP Path from the SQL warehouse connection details.
3. Authenticate with a personal access token, or OAuth / Azure AD.
4. In the navigator, expand the Unity Catalog catalog → schema and pick the gold tables.

Choose the storage mode deliberately:

• DirectQuery sends every interaction to the warehouse, so Unity Catalog’s row/column policies apply on every query. Prefer this for regulated data.
• Import caches into Power BI; governance applies at load time and later changes don’t reach the cache until refresh. Faster, fine for non-sensitive views.

Tableau

Tableau has had a built-in Databricks connector since 2019.4.

1. Connect → To a Server → Databricks.
2. Enter the Server Hostname and HTTP Path from the warehouse connection details.
3. Authenticate with a personal access token or OAuth.
4. Pick the catalog and schema, then the gold tables Kirimana materialised.

Tableau reads Unity Catalog column comments and table descriptions as field descriptions in the data source — so descriptions a steward wrote in the contract’s description: surface as hover tooltips in the workbook. Use a Live connection to keep Unity Catalog policies enforced on every query, or an Extract when caching is acceptable and the data isn’t sensitive.

Qlik

Qlik Sense reaches Databricks through its Databricks connector.

1. Add data → New connection → Databricks.
2. Provide the Server Hostname and HTTP Path, and authenticate.
3. Select the Unity Catalog catalog and schema, then the gold tables.

Qlik caches the table and column metadata on connect, so refresh the connection after Kirimana adds or changes contracts. For row-level security, let Unity Catalog own the rule and have Qlik honour the query result — avoid running two policy models in parallel that can drift.

What the contract metadata gives the BI layer

The reason to read Kirimana-governed gold rather than raw tables is that the governance travels with the data into Unity Catalog:

• Classifications. Each contracted table carries a kiri.classification (public / internal / confidential / restricted). Projected to Unity Catalog, it tells the analyst — and the tool’s own sensitivity handling — how careful to be with a column, not just what it’s called.
• PII flags. Columns flagged as PII in the contract are visible as metadata in the catalog, so a BI author knows a field is personal data before they put it on a public dashboard.
• Ownership. The contract owner is the authoritative “who owns this table”, so a question about a number has an addressee.
• Descriptions. The description: on a contract column materialises as the column comment in Unity Catalog, which tools like Tableau surface as field descriptions — steward-authored context, in the analyst’s hands, without a second copy to maintain.

Because the enforcement lives in Unity Catalog, it applies regardless of which of the three tools an analyst uses. Prefer live / DirectQuery connections for sensitive data so those policies run on every query; reserve caching for views where the data doesn’t need per-query governance.

Troubleshooting

• Connection refused / can’t reach the warehouse. The SQL warehouse may be stopped. Serverless warehouses cold-start on first query; a classic warehouse must be running. Check its state in the workspace.
• “Catalog / schema not found”. Confirm the token’s identity has Unity Catalog grants on the gold catalog and schema, and that your dbt build actually materialised into the catalog you’re browsing.
• Metadata looks stale. BI tools cache the catalog. Refresh the connection after a Kirimana contract change so new classifications and descriptions appear.