Configure dataflow endpoints for Microsoft Fabric OneLake

Important

This page includes instructions for managing Azure IoT Operations components using Kubernetes deployment manifests, which is in preview. This feature is provided with several limitations, and shouldn't be used for production workloads.

See the Supplemental Terms of Use for Microsoft Azure Previews for legal terms that apply to Azure features that are in beta, preview, or otherwise not yet released into general availability.

To send data to Microsoft Fabric OneLake in Azure IoT Operations, you can configure a dataflow endpoint. This configuration allows you to specify the destination endpoint, authentication method, table, and other settings.

Prerequisites

• An instance of Azure IoT Operations
• Microsoft Fabric OneLake. See the following steps to create a workspace and lakehouse.
  • Create a workspace. The default my workspace isn't supported.
  • Create a lakehouse.
  • If shown, ensure Lakehouse schemas (Public Preview) is unchecked.
  • Make note of the workspace and lakehouse names.
• Ensure that service principals can use Fabric APIs.

Assign permission to managed identity

To configure a dataflow endpoint for Microsoft Fabric OneLake, we recommend using either a user-assigned or system-assigned managed identity. This approach is secure and eliminates the need for managing credentials manually.

After the Microsoft Fabric OneLake is created, you need to assign a role to the Azure IoT Operations managed identity that grants permission to write to the Fabric lakehouse.

If using system-assigned managed identity, in Azure portal, go to your Azure IoT Operations instance and select Overview. Copy the name of the extension listed after Azure IoT Operations Arc extension. For example, azure-iot-operations-xxxx7. Your system-assigned managed identity can be found using the same name of the Azure IoT Operations Arc extension.

Go to Microsoft Fabric workspace you created, select Manage access > + Add people or groups.

1. Search for the name of your user-assigned managed identity set up for cloud connections or the system-assigned managed identity. For example, azure-iot-operations-xxxx7 .
2. Select Contributor as the role, then select Add. This gives the managed identity the necessary permissions to write to the Fabric lakehouse. To learn more, see Roles in workspaces in Microsoft Fabric.

Create dataflow endpoint for Microsoft Fabric OneLake

Portal

1. In the operations experience, select the Dataflow endpoints tab.

2. Under Create new dataflow endpoint, select Microsoft Fabric OneLake > New.

   

3. Enter the following settings for the endpoint:

   Setting	Description
   Host	The hostname of the Microsoft Fabric OneLake endpoint in the format onelake.dfs.fabric.microsoft.com.
   Lakehouse name	The name of the lakehouse where the data should be stored.
   Workspace name	The name of the workspace associated with the lakehouse.
   OneLake path type	The type of path used in OneLake. Select Files or Tables.
   Authentication method	The method used for authentication. Choose System assigned managed identity or User assigned managed identity.
   Client ID	The client ID of the user-assigned managed identity. Required if using User assigned managed identity.
   Tenant ID	The tenant ID of the user-assigned managed identity. Required if using User assigned managed identity.

4. Select Apply to provision the endpoint.

Bicep

Create a Bicep .bicep file with the following content.

param aioInstanceName string = '<AIO_INSTANCE_NAME>'
param customLocationName string = '<CUSTOM_LOCATION_NAME>'
param endpointName string = '<ENDPOINT_NAME>'
param workspaceName string = '<WORKSPACE_NAME>'
param lakehouseName string = '<LAKEHOUSE_NAME>'

resource aioInstance 'Microsoft.IoTOperations/instances@2024-11-01' existing = {
  name: aioInstanceName
}
resource customLocation 'Microsoft.ExtendedLocation/customLocations@2021-08-31-preview' existing = {
  name: customLocationName
}
resource oneLakeEndpoint 'Microsoft.IoTOperations/instances/dataflowEndpoints@2024-11-01' = {
  parent: aioInstance
  name: endpointName
  extendedLocation: {
    name: customLocation.id
    type: 'CustomLocation'
  }
  properties: {
    endpointType: 'FabricOneLake'
    fabricOneLakeSettings: {
      // The default Fabric OneLake host URL in most cases
      host: 'https://onelake.dfs.fabric.microsoft.com'
      authentication: {
        // See available authentication methods section for method types
        // method: <METHOD_TYPE>
      }
      oneLakePathType: 'Tables'
      names: {
        workspaceName: workspaceName
        lakehouseName: lakehouseName
      }
    }
  }
}

Then, deploy via Azure CLI.

az deployment group create --resource-group <RESOURCE_GROUP> --template-file <FILE>.bicep

Kubernetes (preview)

Create a Kubernetes manifest .yaml file with the following content.

apiVersion: connectivity.iotoperations.azure.com/v1
kind: DataflowEndpoint
metadata:
  name: <ENDPOINT_NAME>
  namespace: azure-iot-operations
spec:
  endpointType: FabricOneLake
  fabricOneLakeSettings:
    # The default Fabric OneLake host URL in most cases
    host: https://onelake.dfs.fabric.microsoft.com
    authentication:
      # See available authentication methods section for method types
      # method: <METHOD_TYPE>
    oneLakePathType: Tables
    names:
      workspaceName: <WORKSPACE_NAME>
      lakehouseName: <LAKEHOUSE_NAME>

Then apply the manifest file to the Kubernetes cluster.

kubectl apply -f <FILE>.yaml

OneLake path type

The oneLakePathType setting determines the type of path to use in the OneLake path. The default value is Tables, which is the recommended path type for the most common use cases. The Tables path type is a table in the OneLake lakehouse that is used to store the data. It can also be set as Files, which is a file in the OneLake lakehouse that is used to store the data. The Files path type is useful when you want to store the data in a file format that isn't supported by the Tables path type.

Portal

The OneLake path type is set in the Basic tab for the dataflow endpoint.

Bicep

fabricOneLakeSettings: {
  oneLakePathType: 'Tables' // Or 'Files'
}

Kubernetes (preview)

fabricOneLakeSettings:
  oneLakePathType: Tables # Or Files

Available authentication methods

The following authentication methods are available for Microsoft Fabric OneLake dataflow endpoints.

System-assigned managed identity

Before you configure the dataflow endpoint, assign a role to the Azure IoT Operations managed identity that grants permission to write to the Fabric lakehouse:

1. In Azure portal, go to your Azure IoT Operations instance and select Overview.
2. Copy the name of the extension listed after Azure IoT Operations Arc extension. For example, azure-iot-operations-xxxx7.
3. Go to Microsoft Fabric workspace, select Manage access > + Add people or groups.
4. Search for the name of your system-assigned managed identity. For example, azure-iot-operations-xxxx7 .
5. Select an appropriate role, then select Add.

Then, configure the dataflow endpoint with system-assigned managed identity settings.

Portal

In the operations experience dataflow endpoint settings page, select the Basic tab then choose Authentication method > System assigned managed identity.

Bicep

fabricOneLakeSettings: {
  authentication: {
    method: 'SystemAssignedManagedIdentity'
    systemAssignedManagedIdentitySettings: {}
  }
}

Kubernetes (preview)

fabricOneLakeSettings:
  authentication:
    method: SystemAssignedManagedIdentity
    systemAssignedManagedIdentitySettings:
      {}

If you need to override the system-assigned managed identity audience, you can specify the audience setting.

Portal

In most cases, you don't need to specify a service audience. Not specifying an audience creates a managed identity with the default audience scoped to your storage account.

Bicep

fabricOneLakeSettings: {
  authentication: {
    method: 'SystemAssignedManagedIdentity'
    systemAssignedManagedIdentitySettings: {
      audience: 'https://<ACCOUNT>.onelake.dfs.fabric.microsoft.com'
    }
  }
}

Kubernetes (preview)

fabricOneLakeSettings:
  authentication:
    method: SystemAssignedManagedIdentity
    systemAssignedManagedIdentitySettings:
      audience: https://<ACCOUNT>.onelake.dfs.fabric.microsoft.com

User-assigned managed identity

To use user-assigned managed identity for authentication, you must first deploy Azure IoT Operations with secure settings enabled. Then you need to set up a user-assigned managed identity for cloud connections. To learn more, see Enable secure settings in Azure IoT Operations deployment.

Before you configure the dataflow endpoint, assign a role to the user-assigned managed identity that grants permission to write to the Fabric lakehouse.:

1. Go to Microsoft Fabric workspace, select Manage access > + Add people or groups.
2. Search for the name of your user-assigned managed identity.
3. Select an appropriate role, then select Add.

Then, configure the dataflow endpoint with user-assigned managed identity settings.

Portal

In the operations experience dataflow endpoint settings page, select the Basic tab then choose Authentication method > User assigned managed identity.

Enter the user assigned managed identity client ID and tenant ID in the appropriate fields.

Bicep

fabricOneLakeSettings: {
  authentication: {
    method: 'UserAssignedManagedIdentity'
    userAssignedManagedIdentitySettings: {
      clientId: '<ID>'
      tenantId: '<ID>'
      // Optional, defaults to 'https://storage.azure.com/.default'
      // scope: 'https://<SCOPE_URL>' 
    }
  }
}

Kubernetes (preview)

To use a user-assigned managed identity, specify the UserAssignedManagedIdentity authentication method and provide the clientId and tenantId of the managed identity.

fabricOneLakeSettings:
  authentication:
    method: UserAssignedManagedIdentity
    userAssignedManagedIdentitySettings:
      clientId: <ID>
      tenantId: <ID>
      # Optional, defaults to 'https://storage.azure.com/.default'
      # scope: https://<SCOPE_URL>

Here, the scope is optional and defaults to https://storage.azure.com/.default. If you need to override the default scope, specify the scope setting using Bicep or Kubernetes.

Advanced settings

You can set advanced settings for the Fabric OneLake endpoint, such as the batching latency and message count. You can set these settings in the dataflow endpoint Advanced portal tab or within the dataflow endpoint custom resource.

Batching

Use the batching settings to configure the maximum number of messages and the maximum latency before the messages are sent to the destination. This setting is useful when you want to optimize for network bandwidth and reduce the number of requests to the destination.

Field	Description	Required
latencySeconds	The maximum number of seconds to wait before sending the messages to the destination. The default value is 60 seconds.	No
maxMessages	The maximum number of messages to send to the destination. The default value is 100000 messages.	No

For example, to configure the maximum number of messages to 1000 and the maximum latency to 100 seconds, use the following settings:

Portal

In the operations experience, select the Advanced tab for the dataflow endpoint.

Bicep

fabricOneLakeSettings: {
  batching: {
    latencySeconds: 100
    maxMessages: 1000
  }
}

Kubernetes (preview)

fabricOneLakeSettings:
  batching:
    latencySeconds: 100
    maxMessages: 1000

Next steps

To learn more about dataflows, see Create dataflow.