Share via

Confusion on Correct Endpoint Type for Azure ML Model Deployment: Managed Online Endpoint versus Kubernetes Online Endpoint

Marwa Eshra 20 Reputation points
2025-06-18T13:32:44.2066667+00:00

In the course called Deploy a model to a managed online endpoint, the first question in the module assessment was asking:
You've trained a model using the Python SDK for Azure Machine Learning. You want to deploy the model to get real-time predictions**. You want to** manage the underlying infrastructure **used by the endpoint. What kind of endpoint should you create?
**
I answered the first answered as well as copilot, which was "A managed online endpoint" but it showed that when reviewing the answers in the module assessment, it showed that correct answer was "A Kubernetes online endpoint."

This was copilot answer for me asking it about the difference:
"
Possible Explanations for the Answer Difference

  1. Infrastructure Control Requirement:
    • If the question explicitly states that the user wants full control over the underlying infrastructure, then Kubernetes Online Endpoint is correct.
    • If the question only asks for real-time deployment with managed infrastructure, then Managed Online Endpoint is correct.
  2. Exam Context:
    • Some DP-100 questions test practical deployments in large-scale environments where Kubernetes might be preferable.
    • Microsoft documentation generally suggests Managed Online Endpoints for ease of deployment.

"

What is the actual correct answer for this question?
And under what conditions one is preferred over the other?

User's image

This question is related to the following Learning Module

Azure | Azure Training

Answer accepted by question author

  1. VarunTha 14,985 Reputation points Microsoft External Staff Moderator
    2025-06-18T15:56:52.0033333+00:00

    Hi Marwa Eshra,

    Thank you for your insightful question and for your commitment to understanding the nuances of Azure ML model deployment. Your attention to detail is commendable, and I appreciate the opportunity to clarify this important topic for you.

    Correct Answer for Azure ML Model Deployment

    The correct answer to the question regarding the type of endpoint to create for deploying a model to get real-time predictions while managing the underlying infrastructure is:

    • Kubernetes Online Endpoint

    Explanation of Endpoint Types

    1.Managed Online Endpoint

    • Use Case: Best for users who prefer ease of deployment without needing to manage the underlying infrastructure.
    • Infrastructure Control: Limited control over the infrastructure; Azure manages it for you.

    2.Kubernetes Online Endpoint

    • Use Case: Ideal for users who require full control over the underlying infrastructure, such as scaling and configuration. -
    • Infrastructure Control: Provides complete control, allowing for custom configurations and optimizations.

    When to Use Each Endpoint

    1.Choose Managed Online Endpoint when:

    • You want a straightforward deployment process.
    • You do not need to manage the infrastructure.

    2.Choose Kubernetes Online Endpoint when:

    • You need full control over the infrastructure.
    • You are deploying in a large-scale environment where custom configurations are necessary.

    This distinction is crucial for ensuring that your deployment meets your specific needs and operational requirements. If you have any further questions or need additional clarification, please feel free to reach out.

    1 person found this answer helpful.

0 additional answers

Sort by: Most helpful
Most helpful Newest Oldest

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.