The Databricks connector allows you to connect Databricks with One Identity Starling enabling you to take advantage of the features and products available with Starling Connect that complement and enhance Databricks' unified platform for data and Artificial Intelligence (AI). Databricks provides services that allow you to migrate complex and expensive on-premises data systems to cloud data platforms.

Supervisor configuration parameters

To configure the connector, following parameters are required:

• Connector name

• Access Token

• Target URL (Format:- https://databricks_instance/api/2.0/preview/scim/v2)

Supported objects and operations

Users

Table 322: Supported operations for Users

Operation	VERB
Create User	POST
Get User	GET
List Users	GET
Update User	PUT
Delete User	DELETE

Groups

Table 323: Supported operations for Groups

Operation	Operation
Create Group	POST
Get Group	GET
List Groups	GET
Update Group	PUT
Delete Group	DELETE

This section lists the mandatory fields required to create a User or Group:

Users

• userName

Groups

• displayName

User and Group mapping

The user and group mappings are listed in the tables below.

Table 324: User mapping

SCIM Parameter	Databricks parameter
id	id
userName	userName
name.familyName	name.familyName
name.givenName	name.givenName
displayName	displayName
emails[0].value	emails[0].value
emails[0].type	emails[0].type
emails[0].primary	emails[0].primary
active	active
groups[].value	groups[].value
groups[].display	groups[].display
entitlements[].value	entitlements[].value

Group

Table 325: Group mapping

SCIM Parameter	Databricks parameter
id	id
displayName	displayName
members[].value	members[].value
members[].display	members[].display
extension.parentGroups[].value	groups[].value
extension.parentGroups[].display	groups[].display
extension.entitlements[].value	entitlements[].value

Connector limitations

• If User's displayName attribute is present in the request, then the name attributes are not considered. If displayName is not provided, the values from the name attributes would be considered.

• The connector doesn't implement roles for Users. Roles endpoints are not supported at target system and the user can not be assigned with random text to the roles.

• The connector does not implement modifying the entitlements for Users or Groups. Entitlements endpoints are not supported at target system. The User or Group can not be assigned with random text to the entitlements.

• Create User operation can create only Active users in the target application. However, Update User operation can be used to make the user InActive.

• When a user create operation is tried with userName of a deleted user, the target system re-activates the deleted user and returns the details of that deleted user (with any updated values for attributes from the create request) instead of creating a new user with new user id.

• The Databricks REST API supports a maximum of 30 requests per second per workspace. A 429 response status code is displayed if the requests exceed the rate limit.

• A Databricks admin user whose Personal Access Token is used to authorize the target endpoints and who is managed by the Identity Provider (IdP) can be deprovisioned using the IdP, which may cause the SCIM provisioning integration to be disabled.

• No meta information is retrieved for users and groups.

• Target system follows soft delete of resources. Inactivating a user by updating 'active' to 'false' soft-deletes the user and that user cannot be retrieved.

• Group's members will not contain 'type' attribute with values 'User' / 'Group' as the target system does not return the type of the members. Groups in target system can have users and groups as members.

• Delete group request with invalid id returns 204 because of the target API behavior. The target API returns success when a deleted group is tried deleting again.

• List users and list groups APIs are relatively slow. Hence used very minimal set of attributes in the query attributes on these endpoint.