Databricks Unity Catalog

Read Databricks Unity Catalog metadata into Telmai for dataset discovery and monitor recommendations

Telmai reads metadata from Databricks Unity Catalog for dataset discovery, column schema, owners, and one-click monitoring of a workspace's tables.

Unlike a metadata aggregator that spans many databases, Unity Catalog is the Databricks lakehouse's own governance layer. That shapes how the integration behaves: one integration represents one workspace, and a single Databricks connection serves every dataset discovered in it.

This integration applies to Telmai Databricks data connections. To bring catalog metadata to assets from other connection types, use the Actian Data Intelligence integration.

What Telmai reads

Metadata	How Telmai uses it
Catalogs, schemas, tables	Dataset discovery — populates the Catalog Browser
Full column schema	Informs monitor recommendations and asset configuration
Primary and foreign key flags	Read from table_constraints; informs uniqueness and relationship rules
Table owner	Recorded as the dataset's steward
Usage and popularity	30-day read counts, normalized into a usage score that helps prioritize which tables matter
Catalog Explorer links	Deep links back to the table in Databricks

By default Telmai harvests MANAGED_CATALOG and FOREIGN_CATALOG catalogs. Databricks system and internal catalogs, and information_schema, are excluded.

Not yet available

Capability	Status
Lineage	Coming soon
Tags, glossary terms, sensitivity labels	Not read from Unity Catalog. Critical data element tiers come from Telmai's AI classifier instead
Data products	Not supported
Writing data quality results back	Not supported. Write-back is available with Actian Data Intelligence

---

Step 1: Create a personal access token in Databricks

Telmai authenticates to your Databricks workspace with a personal access token (PAT).

1. In your Databricks workspace, go to Settings > Developer > Access tokens.

2. Select Generate new token, give it a descriptive comment (for example, telmai-catalog), and set a lifetime.

3. Copy the token. Databricks displays it once.

Required privileges

The identity behind the token needs Unity Catalog read access — USE CATALOG on each catalog you want Telmai to read.

GRANT USE CATALOG ON CATALOG <catalog_name> TO `<user-or-service-principal>`;

Without USE CATALOG, the catalog and everything beneath it is invisible to Telmai — the sync completes successfully but discovers no datasets.

Usage and popularity scores are collected by querying Databricks system tables through a SQL warehouse. If the token cannot reach a SQL warehouse, discovery still works and usage scores are simply absent.

---

Step 2: Connect to Unity Catalog

In Telmai, go to Administration and select Catalog from the left menu.

Field	Description
Catalog Type	Select Databricks Unity Catalog
Workspace URL	Your Databricks workspace URL, for example https://adb-xxxx.azuredatabricks.net
Personal Access Token	The PAT created in Step 1. Leave blank when updating other fields to keep existing credentials
Sync Interval (hours)	How often Telmai syncs metadata. Accepts values between 1 and 168 hours

Select Save Changes. Creating the integration runs a connection test and triggers an immediate first sync. Sync Now triggers a sync at any time thereafter.

The integration name also labels the source row in the Catalog Browser, so give it a name that reads well there — for example, Databricks Dev Workspace.

Only one catalog can be connected at a time. Remove the existing connection before connecting a different catalog.

Advanced Settings

Advanced Settings are optional key/value settings that fine-tune how Telmai syncs your Unity Catalog integration. All of them have sensible defaults — for a typical setup, leave them empty. Use + Add Setting to add one.

Property	Default	Description
catalogs-include	(empty — all catalogs)	Comma-separated list of catalog names to sync. When set, only these catalogs are harvested. Useful for limiting large workspaces to the catalogs you actually monitor.
catalogs-exclude	(empty)	Comma-separated list of catalog names to skip. Applied after catalogs-include.
catalog-types	MANAGED_CATALOG,FOREIGN_CATALOG	Which Unity Catalog catalog types are synced. By default, Databricks-managed system and internal catalogs are excluded. Set a blank value to disable type filtering entirely.
source-name	(integration name)	Overrides the source label shown in Catalog Browser. By default, the integration's name is used.
usage-warehouse-id	(picked automatically)	ID of the SQL warehouse Telmai uses to query Databricks system tables for usage (popularity) scores. By default, Telmai reuses the warehouse it detects for connection setup. Set this when your workspace has several warehouses and you want sync queries to run on a dedicated one, instead of waking or competing with the warehouse used for data scans.
usage-refresh-hours	0 (refresh on every sync)	Minimum number of hours between usage-score refreshes. Raise this (for example, to 24) to reduce SQL warehouse wake-ups on cost-sensitive workspaces.

---

Step 3: Catalog Browser

Catalog Browser is available in the sidebar under Configure. It is where you browse catalog-discovered datasets and manage which tables are connected for monitoring.

Catalog Connections

A Unity Catalog integration appears as exactly one source — the whole workspace. Datasets from every catalog and schema in that workspace live under it.

• Connected — the source is linked to a Telmai Databricks connection. That one connection serves every dataset in the workspace.

• Not Connected — the source has not yet been linked.

Connecting the source

Select Connect on the source to open a new Databricks connection form. Telmai prefills almost all of it from the workspace — host, port, catalog usage, and the HTTP path looked up live from the workspace's SQL warehouses. In most cases the only value you type is the Databricks token.

Linking the source connects every dataset in the workspace in one action, and datasets discovered by later syncs are linked to the same connection automatically. The catalog and schema are recorded per asset, so a single connection covers them all.

If the token could not read the workspace's SQL warehouses, the HTTP path is not prefilled — enter it manually.

All Sources

The right panel shows all tables discovered from Unity Catalog, grouped into two sections:

• Connected tables — already linked to Telmai assets

• Non connected tables — discovered from the catalog but not yet monitored in Telmai

Each table shows the following columns:

Column	Description
Score	Current data quality score for the table
Details	Number of columns
Last changed	When the table was last updated in the catalog
Why it's recommended	AI-generated explanation of why Telmai recommends monitoring this table

---

Intelligent Recommendations

Telmai uses Unity Catalog metadata to prioritize which tables most need monitoring. The Why it's recommended column surfaces this reasoning for each table, drawing on:

• Column schema and key constraints

• Usage and popularity — how heavily the table has been read over the last 30 days

• Ownership completeness

• Critical data element tiers assigned by Telmai's AI classifier

Select any table to open its Catalog Details, which shows the dataset's catalog metadata, the full AI recommendation with its reasoning and priority, connection status, and suggested timestamp columns.

See Monitor Recommendations for how recommended rules are reviewed and added.

---

Metadata syncs on the interval configured in Administration > Catalog. Use Sync Now to trigger an immediate sync at any time.

Related pages

• Catalog Integration

• Actian Data Intelligence

• Databricks connector

• Monitor Recommendations