PM update: authentication options and network limitations

Author: msakande

articles/machine-learning/concept-model-monitoring.md

@@ -170,6 +170,26 @@ You can use events generated by Azure Machine Learning model monitoring runs to
 
 For example, if the accuracy of your classification model in production dips below a certain threshold, you can use Event Grid to begin a retraining job that uses collected ground truth data. To learn how to integrate Azure Machine Learning with Event Grid, see [Monitor performance of models deployed to production](how-to-monitor-model-performance.md).
 
+## Model monitoring authentication options
+
+Azure Machine Learning model monitoring supports both credential-based and credential-less authentication to the datastore with the collected production inference data from your model. To configure credential-less authentication, follow the steps below:
+
+1) Create a User-Assigned Managed Identity (UAMI) and attach it to your Azure Machine Learning workspace
+2) Grant the UAMI [proper permissions](https://learn.microsoft.com/en-us/azure/machine-learning/how-to-identity-based-service-authentication?view=azureml-api-2&tabs=cli#user-assigned-managed-identity) to access your datastore.
+3) Update the workspace level property `systemDatastoresAuthMode` to `'identity'`.
+
+Or, you can simply add credentials to the datastore where your production inference data is being stored.
+
+To learn more about credential-less authentication with Azure Machine Learning, see [here](https://learn.microsoft.com/en-us/azure/machine-learning/how-to-identity-based-service-authentication?view=azureml-api-2&tabs=cli#user-assigned-managed-identity).
+
+## Model monitoring networking limitations
+
+Azure Machine Learning model monitoring does not support the `AllowOnlyApprovedOutbound` managed virtual network isolation setting. To learn more about managed virtual network isolation in Azure Machine Learning, see [here](https://learn.microsoft.com/en-us/azure/machine-learning/how-to-managed-network?view=azureml-api-2&tabs=azure-cli).
+
+## Model monitoring other limitations
+
+Azure Machine Learning model monitoring has a dependency on `Spark` to compute metrics over large-scale datasets. `MLTable` is not well-supported by `Spark`, and thus it is recommended to avoid using `MLTable` whenever possible with model monitoring jobs. Only basic `MLTable` files are guaranteed to be supported. For complex or custom operations, we recommend using the `Spark` API directly in your code.
+
 ## Related content
 
 - [Model data collection](concept-data-collection.md)