AI-300: Operationalizing Machine Learning and Generative AI SolutionsDemo

By Microsoft · Question Mode

QuestionQ163

Implement machine learning model lifecycle and operations

You manage an Azure Machine Learning workspace by using Python SDK v2.

You must create a compute cluster in the workspace. The cluster must run workloads and handle interruptions properly. First, calculate the workloads’ maximum required compute resources and size the cluster accordingly.

The cluster definition has these properties and values:

  • name="mlcluster1"
  • size="STANDARD_DS3_v2"
  • min_instances=1
  • max_instances=4
  • tier="dedicated"

Compute-resource cost must be minimized both while a workload is active and while the cluster is idle. Changing cluster properties must not change the maximum compute resources available to workloads running on the cluster.

You need to modify the cluster properties to reduce compute-resource cost.

Which properties should you modify?

Select
active
idle
Explanation

Use the low_priority tier for interruptible workloads because it costs less than dedicated compute but is preemptible. Set min_instances to 0 so the cluster can scale down to zero nodes while idle. Retaining max_instances=4 preserves the required maximum workload capacity.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!