Data Modelling Tools for Databricks: What Reads Unity Catalog
What Databricks already draws for free, what five modelling tools read from Unity Catalog, and how to put your own catalog on one canvas.
On this page
Words in this article
Words in this article
- Unity Catalog
- The Databricks layer that organizes tables as catalog › schema › table.
- Lakehouse
- Files in cloud storage that you query like a database.
- Delta Lake
- The table format Databricks uses; constraints need it.
- SQL warehouse
- The Databricks compute that runs SQL queries.
- ER diagram
- A drawing of tables and the keys that link them.
- Informational key
- A key Databricks records but never checks.
- Virtual foreign key
- A link saved only in the DbSchema model, never in the database.
- JDBC driver
- The Java library that lets a tool talk to one kind of database.
- httpPath
- The part of the JDBC URL that names one SQL warehouse.
- DDL
- SQL statements that create or change tables.
The short answer
A data modelling tool draws the tables of a database and lets you plan changes to them. On Databricks, the tables live in Unity Catalog, the layer that organizes them as catalog › schema › table. Pick by what you need: a free first look, the whole catalog in Git, or specialist coverage of certain objects.
A free first look
Catalog Explorer in your workspace draws the keys that tables declare. It costs nothing.
The whole catalog, in Git
DbSchema draws every catalog in one diagram for free. DbSchema Pro keeps the model in Git and writes documentation anyone can open.
Specialist coverage
erwin, ER/Studio, Hackolade and SqlDBM each list Databricks objects on their own pages.
What DbSchema adds on Databricks
Why DbSchema
A lakehouse keeps its data as files in cloud storage and lets you query them like a database. As it grows, it spreads over many catalogs and hundreds of schemas. Catalog Explorer draws the declared keys around the one table you open, so the bigger picture stays in people's heads. DbSchema comes in three editions. Community is free. Pro adds the model file, Git, synchronization and documentation. Architect adds logical design.
Every catalog in one diagram
DbSchema reverse-engineers Unity Catalog. It reads tables and keys through your SQL warehouse, the Databricks compute that runs SQL. Then it draws them. Free in Community.
Links the lakehouse never declared
Draw a virtual foreign key where two tables belong together. Model → Find Virtual Foreign Keys suggests them from matching column names.
A model file in Git (Pro)
The design lives in the .dbs file, DbSchema's own project file. Your team reviews it in Git like code.
Answers without a workspace login (Pro)
One export builds HTML5 documentation with diagrams and column comments. Teammates open it in any browser.
Then, in DbSchema:Reverse-engineerModel in GitRead the DDLSynchronize
What Catalog Explorer draws for free
Every Databricks workspace has a built-in ER diagram, a drawing of tables and the keys that link them. Catalog Explorer shows the primary key and foreign key relationships between tables as a graph[1].
To open it, select a table that has a foreign key. Then click View relationships on its Overview tab[1]. For how data moves between tables, Databricks points you to the lineage graph in Catalog Explorer instead.
The diagram shows only keys that someone declared. A table without declared keys appears with no lines at all. A modelling tool is worth adding when it does what this diagram cannot. It can draw links nobody declared, or show the whole catalog at once. It can keep a model you edit offline and review in Git. It can document the schema for people outside the workspace.
Modelling tools that read Unity Catalog
Five tools name Databricks support on their own pages. We read each page on 15 November 2026. The ER/Studio page refused access that day, so its row comes from our check on 10 October 2026.
| Tool | What its page says it reads | Model kept in | Runs on |
|---|---|---|---|
| DbSchema | Catalogs, schemas and tables | A .dbs file in your Git repository | Windows, macOS, Linux |
| erwin Data Modeler | Catalogs, tables, columns, partitions, views, functions, CTAS | Not stated | Not stated |
| ER/Studio Data Architect | Databricks as a core platform | A repository with Team Server | 64-bit Windows 10 and 11 |
| Hackolade Studio | Delta Lake tables, Hive complex and user-defined types | Not stated | Not stated |
| SqlDBM | Schemas, tables, views, partitions, keys, check constraints | SqlDBM projects | Browser |
DbSchema keeps the model as a file in your own Git repository. The other vendors' Databricks pages do not state this.
Each page adds its own strength:
- erwin Data Modeler lists CTAS among its Databricks objects. CTAS means CREATE TABLE AS SELECT, a table built from a query[2].
- ER/Studio Data Architect reverse engineers from the database and forward engineers DDL. Forward engineering writes DDL, the SQL that creates or changes tables, from the model[3].
- Hackolade Studio models Hive complex types, the nested types such as arrays, maps and structs. It writes HiveQL scripts, the SQL dialect of Hive[4].
- SqlDBM reads the DDL from a Databricks cluster, the compute that runs notebooks and jobs. Teams work on parallel branches of a SqlDBM project[5].
Keys in Databricks are informational
Databricks records primary keys, foreign keys and unique constraints, but it never checks them. That is what an informational key means[6]. NOT NULL and CHECK constraints are different: Databricks checks them before it adds a row. All constraints need Delta Lake, the table format Databricks uses.
So a diagram of declared keys shows intent, not a guarantee. This limit applies to every tool in the table. Where nobody declared the keys, add virtual foreign keys in DbSchema. A virtual foreign key is a link saved only in the DbSchema model. DbSchema never creates it in Databricks.
Two DbSchema Pro tools follow a virtual foreign key like a real one. The Query Builder writes SQL from the tables you click. The Data Explorer shows the rows of one table and opens the related rows of the next.
Three problems come up on a first connection:
What it means
The JDBC URL has no httpPath. We saw this with the open-source Databricks JDBC driver 3.4.2 on 10 October 2026, before any network request.
Fix
Copy the URL from the warehouse's Connection details tab, with its httpPath.
What it means
The SQL warehouse is stopped and starts slowly.
Fix
Start the warehouse in the workspace first, or allow a generous timeout.
What it means
The tables declare no keys.
Fix
Add virtual foreign keys, as described above.
How DbSchema connects and keeps the model
DbSchema talks to Databricks through the Databricks JDBC driver. A JDBC driver is the Java library that lets a tool talk to one kind of database. The driver connects to one SQL warehouse.
Driverload once
databricks-jdbc-<version>.jar
com.databricks.client.jdbc.Driver
The download is a zip. Extract it and load the .jar in the JDBC Drivers Manager.
JDBC URLkeep httpPath
jdbc:databricks://HOST:443/default;
ssl=1;httpPath=...;AuthMech=3;
UID=token;PWD=...
httpPath selects the SQL warehouse. Without it, the driver reaches the workspace but no compute to run SQL.
Logintoken or OAuth
UID=token
PWD=<personal access token>
The user is the literal word token. Databricks recommends OAuth for production.
The model lives apart from the warehouse, in the .dbs file. You can edit it offline, with no connection open. Edits change only the file. Schema Synchronization compares the model with the live warehouse and lists each difference. DbSchema then shows the DDL for the changes you pick, and runs it only when you say so.
Documentation without workspace credentials
Analysts and auditors often need schema answers but have no workspace login. One export from the DbSchema model builds HTML5 documentation: searchable diagrams, table definitions and column comments in one page. Comments on tables and columns show as tooltips when you point at them. The Community Edition documents diagrams of up to 12 tables; Pro documents any diagram.
Which tool fits which team
| What the team needs | Where it sits |
|---|---|
| A free look at declared keys | Catalog Explorer, already in the workspace |
| CTAS tables and partitions in the model | erwin Data Modeler |
| Nested Hive types such as arrays and maps | Hackolade Studio |
| A browser tool with project branches | SqlDBM |
| Every catalog in one diagram (free), plus a model file in Git and documentation (Pro), on Windows, macOS or Linux | DbSchema |
Catalog Explorer is the baseline. Each other tool covers a job the workspace does not.
How to open a Databricks SQL warehouse in DbSchema
-
Install DbSchema Community
Download DbSchema for Windows, macOS or Linux. No account is needed.
-
Load the Databricks driver
Download the Databricks JDBC driver zip and extract it. In DbSchema, choose Database → Manage → JDBC Drivers Manager, pick Databricks and press Upload Driver Jar(s).
-
Copy the JDBC URL
Open the SQL warehouse's Connection details tab and copy the URL with its httpPath. Start the warehouse.
-
Connect
Choose the Method Paste a JDBC URL, paste the URL, and sign in with a personal access token or OAuth.
-
Reverse-engineer the catalog
In the Reverse Engineer the Database dialog, tick the schemas to import and press Import. Catalogs, schemas and tables appear as a diagram.
Every download starts with a 15-day trial of all features. Afterwards, Community keeps connecting, reverse-engineering, drawing diagrams and running SQL, for every database. Download DbSchema and see your Unity Catalog in one diagram.
Frequently asked questions
Does Databricks have a built-in ER diagram?
Yes. Catalog Explorer draws it for any table that has a foreign key. Open it with View relationships on the table's Overview tab.
Are primary and foreign keys enforced in Databricks?
No. Only NOT NULL and CHECK constraints are checked when a row is written.
Can I model Unity Catalog offline and keep the model in Git?
Yes, with DbSchema Pro. You edit the .dbs file offline and commit it to Git like any other file.
Sources
- Databricks: Entity relationship diagram, read 15 November 2026
- The erwin Data Modeler Databricks Support Summary on the Quest bookshelf, read 15 November 2026
- The ER/Studio Data Architect Technical Specifications on the IDERA wiki, read 10 October 2026
- The Hackolade page on Databricks data modeling, read 15 November 2026
- The SqlDBM page on SqlDBM and Databricks, read 15 November 2026
- Databricks: Constraints on Databricks, read 15 November 2026
Put Unity Catalog on one canvas
DbSchema connects to a Databricks SQL warehouse through the JDBC driver and draws catalogs, schemas and tables as ER diagrams. Connecting, reverse-engineering and the diagrams are in the free Community Edition.