Key takeaways

  • Creating a catalog requires the CREATE CATALOG privilege on the Unity Catalog metastore.
  • If the metastore has no storage of its own, a standard catalog needs a MANAGED LOCATION inside an external location, or Default Storage in a serverless workspace.
  • A catalog bound to specific workspaces is refused everywhere else, even for users who hold grants on it.

To create a catalog in Databricks, you need the CREATE CATALOG privilege on your Unity Catalog metastore. Then either click Catalog > Create catalog in Catalog Explorer or run CREATE CATALOG IF NOT EXISTS my_catalog; in a notebook or the SQL editor. Every step below comes from Databricks' own Create catalogs page and the pages it links to, as they read on November 21, 2026.

This guide is for data engineers and admins setting up Unity Catalog. When you're done you will have a catalog with a storage location and the right workspace access, plus a first schema to put tables in.

A catalog is the top level of Unity Catalog's three-level namespace, catalog.schema.table. It holds schemas, and schemas hold tables, views, volumes, models and functions. Databricks suggests making each catalog a unit of isolation, such as one for production and one for development, or one for customer data and one for everything else.

What you need before you create a catalog

From the requirements on the docs page:

  • The CREATE CATALOG privilege on the metastore. Metastore admins and account admins can grant it.
  • Compute that supports Unity Catalog. From a notebook, that means Databricks Runtime 11.3 or above with a Unity Catalog-compliant access mode. SQL warehouses always work.
  • A place to store managed tables. If your metastore has no storage location of its own, you must give the catalog one. To set one, you need the CREATE MANAGED STORAGE privilege on an external location that covers the path.

That last point catches a lot of people. According to the managed storage page, workspaces where Unity Catalog was turned on automatically come without metastore-level storage. Unity Catalog is turned on automatically for every workspace created after November 8, 2023, per Databricks. So on a newer account, a bare CREATE CATALOG with no location may fail.

Serverless workspaces can use Default Storage

Default Storage is object storage that Databricks manages for you. Serverless workspaces use it for the default catalog that comes with the workspace, and you can put new catalogs there too, with no external location to set up. There are limits:

  • You can only create default storage catalogs in serverless workspaces.
  • By default, only the workspace that created the catalog can reach it. Other workspaces can be given access, but they have to use serverless compute.
  • Classic compute can't read or write data in default storage.

How to create a catalog in Catalog Explorer

  1. Log in to a workspace attached to the metastore.
  2. Click Catalog in the sidebar, then Catalogs under Quick access.
  3. Click Create catalog.
  4. Type a Catalog name and pick a Type: Standard for ordinary Unity Catalog data, Foreign to mirror an outside database, or Shared for a catalog built from a share someone sent you.
  5. For a standard catalog, set a Storage location. Databricks calls this optional but strongly recommended, and required when the metastore has no storage. You can use an external location or a subpath of one, or click Create a new external location. In a serverless workspace you can pick Use default storage instead.
  6. Click Create.
  7. On the confirmation dialog that follows, click Configure catalog, which Databricks recommends. If you click View catalog instead, every workspace on the metastore gets access and all account users get BROWSE.
  8. On Workspaces, keep All workspaces have access, or clear it and use Assign to workspaces to limit the catalog. The current workspace has to stay on the list. Use Manage Access Level to make a workspace read-only.
  9. On Permissions, click Grant to give users, groups or service principals privileges.
  10. Under Metadata, add tags and a comment if you want, then click Save.

Every new catalog starts with two schemas, default and information_schema.

How to create a catalog with SQL

Run this in a notebook or the SQL editor. Items in square brackets are optional:

CREATE CATALOG [ IF NOT EXISTS ] <catalog-name>
  [ MANAGED LOCATION '<location-path>' ]
  [ COMMENT <comment> ];

The simplest version:

CREATE CATALOG IF NOT EXISTS example;

With its own storage location and a comment, which is the setup Databricks recommends:

CREATE CATALOG IF NOT EXISTS finance
  MANAGED LOCATION 's3://depts/finance'
  COMMENT 'Finance team data';

The path has to be an external location you've already defined, or a subpath of one, and the whole string can be at most 150 characters. These examples come from the AWS docs, which use s3:// paths. Databricks publishes separate versions of the docs for Azure and Google Cloud. Without IF NOT EXISTS, the CREATE CATALOG reference says the statement throws an error if a catalog with that name exists.

The reference lists a few more options. DEFAULT COLLATION needs Databricks Runtime 17.1 or above. RETAIN DROPPED FOR sets how long dropped managed tables can be recovered. It is in Public Preview and needs Runtime 17.3 or above:

CREATE CATALOG customer_cat RETAIN DROPPED FOR 30 DAYS;

On Databricks Runtime 18.1 and above you can move a catalog's managed location later. Only new tables and volumes go to the new path. Existing ones stay where they are:

ALTER CATALOG finance
  SET MANAGED LOCATION 's3://depts/finance-v2';

How to create a catalog with the Databricks CLI

The CLI reference lists --comment, --storage-root, --connection-name, --provider-name, --share-name and --json as options:

databricks catalogs create sales --comment "Sales data catalog"

Add --storage-root with your path to set the managed location. There's also a REST endpoint, POST /api/2.1/unity-catalog/catalogs.

How to create a catalog with Terraform

Databricks points to the databricks_catalog resource in its Terraform provider. The provider's documentation gives this example:

resource "databricks_catalog" "sandbox" {
  name    = "sandbox"
  comment = "this catalog is managed by terraform"
  properties = {
    purpose = "testing"
  }
}

Set storage_root for the managed location. You need it unless the metastore has storage. Set isolation_mode = "ISOLATED" to bind the catalog to the current workspace. Two more things to know: the resource only works with a workspace-level provider, and Terraform deletes the auto-created default schema so the catalog can be destroyed cleanly.

Foreign and shared catalogs

A foreign catalog mirrors a database in another system through Lakehouse Federation, so you can run read-only queries on it from Databricks. You need a connection first, plus ownership of it or CREATE FOREIGN CATALOG on it. Compute has to be Runtime 13.1 or above, or a Pro or Serverless SQL warehouse.

CREATE FOREIGN CATALOG postgresql_catalog
  USING CONNECTION postgresql_connection
  OPTIONS (database = 'postgresdb');

MySQL doesn't need a database name. To mirror a catalog in another Databricks workspace, use catalog '<name>' in place of database. A shared catalog comes from a share you have received. You need USE PROVIDER on the metastore or ownership of the provider:

CREATE CATALOG customer_cat USING SHARE cdc.vaccinedata;

Give people access and create a schema

Users start with no access to data in a metastore, so assign privileges once the catalog exists. To work with anything inside it, a user needs USE CATALOG. The privileges page shows the grant pattern:

GRANT USE CATALOG ON CATALOG finance TO `finance-team`;
GRANT CREATE SCHEMA ON CATALOG finance TO `finance-team`;

Then create a schema. You need USE CATALOG and CREATE SCHEMA on the parent catalog. The Create schemas page gives this syntax:

CREATE SCHEMA IF NOT EXISTS finance.reporting
  COMMENT 'Monthly reporting tables';

You can give the schema its own MANAGED LOCATION too. Unity Catalog stores a managed table in the schema's location first, then the catalog's, then the metastore's.

Make the new catalog the default

If you want queries without a catalog name to use the new catalog, a workspace admin can open Admin Settings > Advanced, enter the name in Default catalog for the workspace and click Save. Then restart SQL warehouses and clusters. Databricks warns that this can break existing code that relies on the old default. For a single session, run USE CATALOG finance;. To check what you're on, run SELECT current_catalog();. Details are on the default catalog page.

Common problems when creating a catalog

Permission denied on CREATE CATALOG. You don't hold the privilege on the metastore. A metastore admin or account admin can run SHOW GRANTS ON METASTORE to check, then grant it:

GRANT CREATE CATALOG ON METASTORE TO `data-engineers`;

You don't name the metastore, because Databricks assumes the one attached to your workspace. A foreign catalog also needs CREATE FOREIGN CATALOG on the connection.

An error about storage when you leave out the location. The metastore has no storage root, so the catalog needs its own. Add MANAGED LOCATION with a path inside an external location where you hold CREATE MANAGED STORAGE. If that external location doesn't exist yet, create it or ask an admin to. In a serverless workspace, Use default storage is the other way out.

The catalog exists but doesn't appear in another workspace. Look at its workspace bindings. Once a catalog is bound, every other workspace is refused, even for users who have grants on it. The workspace catalog that comes with a new workspace, and default storage catalogs, are bound to their own workspace from the start. Catalog owners, metastore admins and users with MANAGE can fix this on the catalog's Workspaces tab with Assign to workspaces, or with the CLI, as the binding page shows:

databricks workspace-bindings update-bindings catalog finance \
  --json '{"add": [{"workspace_id": <workspace-id>, "binding_type": "BINDING_TYPE_READ_WRITE"}]}'

Metastore admins and catalog owners see unbound catalogs grayed out, with nothing inside them visible.

The catalog can't be found from a cluster. Clusters that don't support Unity Catalog fall back to hive_metastore as their default catalog. Use a SQL warehouse or a cluster in standard or dedicated access mode.

Databricks is also in the news this week: on November 20, 2026, The Wall Street Journal reported it is in talks to raise money at about $225 billion, covered here.

Frequently asked questions

How do I create a schema in Databricks?

With USE CATALOG and CREATE SCHEMA on the parent catalog, run CREATE SCHEMA IF NOT EXISTS my_catalog.my_schema; or click Create schema on the catalog in Catalog Explorer. You can add a MANAGED LOCATION if you hold CREATE MANAGED STORAGE on the external location.

What is the difference between a catalog and a schema in Databricks?

A catalog is the top level of Unity Catalog's catalog.schema.table namespace and usually marks a unit of isolation, such as production or development data. A schema sits inside a catalog and holds the tables, views, volumes, models and functions.

How do I create a view in Databricks?

Run CREATE VIEW my_catalog.my_schema.my_view AS SELECT ... in the SQL editor or a notebook. You need USE CATALOG on the catalog, USE SCHEMA and CREATE TABLE on the schema, and SELECT on the tables the view reads.

Why can't I see my Databricks catalog in another workspace?

The catalog is probably bound to specific workspaces. A catalog owner, metastore admin or user with MANAGE can add the workspace on the catalog's Workspaces tab or with the databricks workspace-bindings update-bindings CLI command.

Sources

  1. Create catalogsDatabricks
  2. CREATE CATALOG (SQL reference)Databricks
  3. Specify a managed storage location in Unity CatalogDatabricks
  4. Default storage in DatabricksDatabricks
  5. Workspace-catalog bindingDatabricks
  6. Manage the default catalogDatabricks
  7. What are catalogs in Databricks?Databricks
  8. Create schemasDatabricks
  9. Create and manage viewsDatabricks
  10. Manage privileges in Unity CatalogDatabricks
  11. Unity Catalog privileges referenceDatabricks
  12. Get started with Unity CatalogDatabricks
  13. catalogs command group (Databricks CLI)Databricks
  14. databricks_catalog resourceDatabricks