Exam DP-100 topic 4 question 32 discussion

Actual exam question from Microsoft's DP-100

Question #: 32
Topic #: 4

You train and register a machine learning model. You create a batch inference pipeline that uses the model to generate predictions from multiple data files.
You must publish the batch inference pipeline as a service that can be scheduled to run every night.
You need to select an appropriate compute target for the inference service.
Which compute target should you use?

A. Azure Machine Learning compute instance
B. Azure Machine Learning compute cluster
C. Azure Kubernetes Service (AKS)-based inference cluster
D. Azure Container Instance (ACI) compute target

Show Suggested Answer

Suggested Answer: B 🗳️

by claps92 at Sept. 9, 2022, 7:49 a.m.

Comments

Submit Cancel

BTAB

Highly Voted 2 years, 4 months ago

I think the confusion here is real-time vs batch. A real-time inference pipeline should use AKS. This is a BATCH inference pipeline. The answer is B.

upvoted 10 times

manualrg

2 years, 4 months ago

I agree the answer is B. Indeed, in the example notebook: Create a Batch Inferencing Service https://github.com/MicrosoftLearning/mslearn-dp100/blob/main/10%20-%20Create%20a%20Batch%20Inferencing%20Service.ipynb Existing compute cluster resource is checked and set as pipeline compute target: try: # Check for existing compute target inference_cluster = ComputeTarget(workspace=ws, name=cluster_name) ... parallel_run_config = ParallelRunConfig( source_directory=experiment_folder, entry_script="batch_diabetes.py", mini_batch_size="5", error_threshold=10, output_action="append_row", environment=batch_env, compute_target=inference_cluster, node_count=2)

upvoted 2 times

...

avinyc

Most Recent 5 months ago

Selected Answer: B

Azure Machine Learning compute clusters are designed to handle batch workloads efficiently, which aligns with the requirement of processing multiple data files

upvoted 1 times

...

2 years, 5 months ago

I think the answer B is indeed correct! There is no wording about a "production-level" or "heavy-workload" deployment (this would always hint to AKS), and according to the documentation: https://learn.microsoft.com/en-us/azure/machine-learning/concept-compute-target a compute-cluster supports job-scheduling aka scheduling to run every night, and scales up and down depending on the configuration.

upvoted 3 times

...

silva_831

2 years, 6 months ago

Yes, Answer should be AKS

upvoted 1 times

...

JTWang

2 years, 7 months ago

Answer should be AKS. Compute targets for inference: 1.Local web service 2.Azure Machine Learning endpoints 3.Azure Machine Learning Kubernetes 4.Azure Container Instances (SDK/CLI v1 only) https://learn.microsoft.com/en-us/azure/machine-learning/concept-compute-target

upvoted 3 times