Path8.Mod1.b - Intro to DevOps Principles for ML - Trigger with Azure ML Events

Question 1

The Microsoft.MachineLearningServices.<Event Types> for Azure Machine Learning in EventGrid that’s raised when an ML Experiment completes

Answer 1

RunCompleted

Question 2

The Microsoft.MachineLearningServices.<Event Types> for Azure Machine Learning in EventGrid that’s raised when an ML Model is registered to a workspace

Answer 2

ModelRegistered

Question 3

The Microsoft.MachineLearningServices.<Event Types> for Azure Machine Learning in EventGrid that’s raised when an Inference Service Deployment is completed

Answer 3

ModelDeployed

Question 4

The Microsoft.MachineLearningServices.<Event Types> for Azure Machine Learning in EventGrid that’s raised when a data drift detection job for two datasets is completed

Answer 4

DatasetDriftDetected

Question 5

The Microsoft.MachineLearningServices.<Event Types> for Azure Machine Learning in EventGrid that’s raised when a run status changed

Answer 5

RunStatusChanged

Question 6

Permissions necessary to use Event Grid and to subcribe to ML Events with your Azure ML Workspace

Answer 6

Like most anything, you need Contributor or Owner

Question 7

Tools used for subscribing to ML Events

Answer 7

Azure Portal, Powershell or Azure CLI

Question 8

TEWs ET PO Ig FC

Five Best Practices for consuming ML Events

Answer 8

• Check the Topic of the Message to ensure it’s coming from the expected Workspace
• Check the EventType to ensure it’s one you can process
• Check etag and sequencer fields to ensure processing order (messages can arrive out of order)
• Ignore fields you don’t understand (new features will break you otherwise)
• Failed or cancelled ML operations will not trigger an Event; consider this when developing your event handlers

Question 9

F LA EH DFP GW (AZ-900)

Options for Event Handler implementation

Answer 9

• Azure Functions
• Azure Logic Apps
• Azure Event Hubs
• Azure Data Factory Pipeline
• Generic webhooks

Question 10

CLI code for subscribing to an ML Event
- What --endpoint and --endpoint-type represent
- Two datapoints the --source-resource-id needs
- The one datapoint the --endpoint needs

Answer 10

• --endpoint and --endpoint-type indicate the Event Handling mechanism
• The EventGrid subscription needs
  – name
  – source (source-resource-id)
  – endpoint type
  – endpoint
  – included-event-types: the events you want to subscribe the handler to
  – Has optional filter by subject start string
• --source-resource-id needs subscription ID and Workspace name
• --endpoint needs subscription ID

// Select the Azure subscription that contains the workspace
az account set --subscription "<name or ID of the subscription>"

// Subscribe to the machine learning workspace. This uses EventHub
az eventgrid event-subscription create 
  --name {eventGridFilterName} 
	--source-resource-id /subscriptions/{subId}/resourceGroups/{RG}/providers/Microsoft.MachineLearningServices/workspaces/{wsName} 
	--endpoint-type eventhub 
	--endpoint /subscriptions/{SubID}/resourceGroups/TestRG/providers/Microsoft.EventHub/namespaces/n1/eventhubs/EH1 
	--included-event-types Microsoft.MachineLearningServices.ModelRegistered 
	--subject-begins-with "models/mymodelname"