# Dataset Management

Dataset Management lets you review the object types a source collects and control how the source aggregates those objects into SailPoint. Use Dataset Management to view and manage resource attributes, datasets and their resources, enable dataset aggregation schedules, perform manual aggregation, and view aggregation history.

> [!NOTE]
>
> Datasets are also available in SailPoint Agentic Fabric. Contact your Customer Success team and refer to [SailPoint Agentic Fabric](<https://documentation.sailpoint.com/saas/help/agentic_fabric/index.html>) for more information.

For more information, refer to [Managing Datasets and Resources](<https://documentation.sailpoint.com/saas/help/sources/datasets.html>) for datasets, resources, owner correlation, and aggregation.

## Datasets

A *dataset* is a named group of one or more related resources that the source aggregates together.

When aggregation runs for a dataset, the source collects all resources in that dataset from the managed system in one aggregation job. Activate or deactivate aggregation at the dataset level, not for individual resources.

> [!NOTE]
>
> Each resource belongs to exactly one dataset for that source. The source defines which resources belong to each dataset. You can enable aggregation and review associated resources, but you cannot create or delete datasets or move resources between them.

### Dataset Aggregation

Configure aggregation on the **Dataset Aggregations** tab for each dataset. Configure each dataset separately. During aggregation, your owner correlation rules are automatically applied to link machine identities to their human owners. For more information, refer to [Configuring Dataset Aggregations](<https://documentation.sailpoint.com/saas/help/sources/datasets.html#configuring-dataset-aggregations>).

## Resources

A resource defines an object type from the managed system. It specifies the kind of object, the attributes the source collects, and how the source identifies and displays each object. Resources form the building blocks of the data model.

In the UI, each resource appears under **Dataset Management** \> **Resources** with its schema. A resource does not run aggregation by itself. The dataset that owns the resource collects it during aggregation.

For more information, refer to [Managing Resources](<https://documentation.sailpoint.com/saas/help/sources/datasets.html#managing-resources>).

### Owner Correlation

Configure owner correlation on each resource to link aggregated objects to human owners. If you do not configure owner correlation, or if the mapping produces no match, an owner is not assigned. For more information, refer to [Configuring Owner Correlation](<https://documentation.sailpoint.com/saas/help/sources/datasets.html#configuring-owner-correlation>).

## Supported Datasets

Go to **Dataset Management \> Datasets** on your Databricks source to view the datasets the connector aggregates. Select a dataset to review its associated resources and configure aggregation.

> [!NOTE]
>
> Configure aggregation for each dataset separately. For more information, refer to [Dataset Aggregation](<https://documentation.sailpoint.com/saas/help/sources/datasets.html#configuring-dataset-aggregations>).

| Dataset name | Associated resources | Dataset ID |
| --- | --- | --- |
| **Databricks Mosaic AI Agents** | Databricks Mosaic AI Agent | `databricks:mosaic` |

For schema attribute definitions, refer to [Databricks Mosaic AI Agent Attributes](<#databricks-mosaic-ai-agent-attributes>).

Before you aggregate Mosaic AI agents, enable agent collection on the **Dataset Aggregations** tab for the dataset in **Dataset Management** \> **Datasets**. For more information, refer to [Dataset Aggregation](<https://documentation.sailpoint.com/saas/help/sources/datasets.html#configuring-dataset-aggregations>).

## Supported Resources

For more information, refer to [Managing Resources](<https://documentation.sailpoint.com/saas/help/sources/datasets.html#managing-resources>).

> [!NOTE]
>
> When you enable dataset aggregation for a selected dataset, all resources in that dataset are collected at once. Activate or deactivate aggregation at the dataset level, not for individual resources.

Under **Dataset Management** \> **Resources**, the Databricks source lists each object type that is aggregated. Select a resource to review its schema on the **Schema** tab.

| Resource name | Resource type | Resource ID |
| --- | --- | --- |
| **[Databricks Mosaic AI Agent](<#databricks-mosaic-ai-agent-attributes>)** | Agent | `databricks:mosaic-agent` |

## Databricks Mosaic AI Agent Attributes

Schema attributes for the **Databricks Mosaic AI Agent** resource (`databricks:mosaic-agent`).

| Attribute name | Description | Type | Entitlement | Multi-Valued |
| --- | --- | --- | --- | --- |
| ACLAssignments | Principal ACL assignments as a stringified JSON array of {principalType, principal, permissionLevel}; populated only when ACL aggregation is enabled | string | No | No |
| agentIdentifier | Composite resource identity: agentName@workspaceId | string | No | No |
| aiGateway | AI gateway settings as stringified JSON (usage\_tracking\_config, inference\_table\_config) | string | No | No |
| config | Entire serving endpoint config as a stringified JSON object, verbatim (includes served\_entities and served\_models) | string | No | No |
| creationTimestamp | Creation time as ISO-8601 UTC (converted from Unix epoch millis) | string | No | No |
| creator | Email or identifier of the user who created the agent | string | No | No |
| creatorDisplayName | Display name of the creator | string | No | No |
| creatorKind | Kind of creator (for example User or ServicePrincipal) | string | No | No |
| description | Description of the agent | string | No | No |
| id | Unique identifier of the serving endpoint (agent) | string | No | No |
| lastUpdatedTimestamp | Last-updated time as ISO-8601 UTC (converted from Unix epoch millis) | string | No | No |
| name | Name of the agent (serving endpoint) | string | No | No |
| permissionLevel | Caller's permission level on the agent (CAN\_MANAGE \| CAN\_QUERY \| CAN\_VIEW) | string | No | No |
| spBusinessApplication | Business application identifier for Mosaic AI agents; fixed value databricks\_mosaic\_ai\_agents | string | No | No |
| spDescription | Agent description; mirrors the description attribute value | string | No | No |
| spStatus | Normalized agent status derived from state. Aggregation: ready READY -\> Active; otherwise Inactive (any NOT\_READY, or missing/unreadable state). Enable Agent only: ready NOT\_READY + config\_update IN\_PROGRESS without system\_update\_failure is reported Active (warm-up), reconciled by the next aggregation | string | No | No |
| state | Operational state as a stringified JSON (ready, config\_update, suspend, system\_update\_failure) | string | No | No |
| tags | Endpoint tags as a stringified JSON array of key/value objects | string | No | No |
| task | Task type (for example agent/v1/responses) | string | No | No |
| workspaceId | ID of the Databricks workspace containing this agent | string | No | No |
| workspaceName | Name of the Databricks workspace containing this agent | string | No | No |