Deploy a scikit-learn model on Vertex AI with minimum cost and high availability?

PMLE By islever Editorial Last updated

Short answer

B Deploy an online Vertex AI prediction endpoint. Set the max replica count to 100

Question

You need to deploy a scikit-leam classification model to production. The model must be able to serve requests 24/7, and you expect millions of requests per second to the production application from 8 am to 7 pm. You need to minimize the cost of deployment. What should you do?
  1. Deploy an online Vertex AI prediction endpoint. Set the max replica count to 1
  2. Deploy an online Vertex AI prediction endpoint. Set the max replica count to 100 Correct Answer
  3. Deploy an online Vertex AI prediction endpoint with one GPU per replica. Set the max replica count to 1
  4. Deploy an online Vertex AI prediction endpoint with one GPU per replica. Set the max replica count to 100

Key Takeaways

This question tests autoscaling configuration and GPU necessity for model serving; the common trap is assuming GPUs help with scikit-learn or that a single replica can handle millions of requests per second.

The correct answer is to deploy an online Vertex AI prediction endpoint with max replica count set to 100 and no GPU. Community consensus confirms that scikit-learn does not support GPU, so adding GPUs only increases cost without performance benefits.

Common mistake: Choosing options with GPUs (C or D) is the most common mistake, because many candidates think GPUs accelerate all ML serving. However, scikit-learn does not support GPU inference, so GPUs add cost without improving throughput.

Expert Analysis

Why the Answer Is Correct

Option B is correct because it provides an online Vertex AI prediction endpoint with a maximum of 100 replicas, allowing the service to autoscale to handle millions of requests per second during peak hours (8 am to 7 pm) while scaling down during off-peak times to minimize cost. Scikit-learn is a CPU-only framework; as noted in comments [2] and [3], GPU support is not available, so adding GPUs would be wasteful. Thus, a high max replica count with CPU-only instances is the best balance of scalability and cost efficiency.

Why the Other Options Are Wrong

Option A is insufficient because a max replica count of 1 means the endpoint can never scale beyond a single instance, which cannot handle millions of requests per second. Options C and D both include GPUs, which are not supported or beneficial for scikit-learn models, as confirmed by community comments [2] and [3]. Option D also adds unnecessary GPU costs with 100 replicas, while Option C has both the GPU problem and insufficient scaling capacity.

Community Comment Notes

Community comments overwhelmingly favor B, with [1] and [5] highlighting that 100 replicas are needed for massive scale, while [2] and [3] emphasize that scikit-learn has no GPU support. Comment [4] suggests that max replica count of 1 might minimize cost but ignores the throughput requirement, a point that reinforces why B is the correct choice. Taken together, the comments show that the key is to scale horizontally with CPU instances, not to depend on GPU acceleration.

Official Reference

These links open the official vendor documentation in a new tab.

Exam Strategy

When answering deployment questions, always check the framework's hardware support and the workload's expected traffic volume. For scikit-learn, never choose GPU-enabled endpoints, and ensure the max replica count is high enough to handle peak loads through horizontal autoscaling.

Community Votes

B
100%

100% of anonymous learners picked answer B. Votes are pick records left by other test-takers — they are not the verified answer.

Discussion 6 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 6

  1. pikachu007 👍 5 Answered B
    B. Deploy an online Vertex AI prediction endpoint. Set the max replica count to 100: This option provides a higher number of replicas (100) to handle the expected high volume of requests during peak hours. While it might result in increased costs, it provides the necessary scalability to manage the incoming traffic efficiently. During non-peak hours, you can consider scaling down the replicas to reduce costs, as Vertex AI allows dynamic scaling based on demand.
  2. BlehMaks 👍 2 Answered B
    scikit-learn doesn't support GPU https://scikit-learn.org/stable/faq.html#will-you-add-gpu-support
  3. 36bdc1e 👍 2
    B we don't need GPU for scikit-learn
  4. AzureDP900 👍 1
    Option A (Deploying an online Vertex AI prediction endpoint. Set the max replica count to 1) is still a good choice for minimizing costs. By setting the max replica count to 1, you are allowing Vertex AI to scale up or down based on load, which means that during off-peak hours, you won't be paying for unnecessary instances.
  5. b1a8fae 👍 1
    B. scikit-learn -> no need for GPU max number of replicas -> 1 is too little if we are serving online predictions at such a massive scale (millions per second)
Show 1 more comment
  1. pinimichele01 👍 1 Answered B
    see pikachu007

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

How to Minimize Development Time for Weekly TensorFlow Retraining in Vertex AI?

S-grade

How to Handle Categorical Features in BigQuery ML for Fast Deployment?

S-grade

How to preprocess BigQuery data for TensorFlow?

S-grade

How to Track Model and Prediction Lineage in Vertex AI?

S-grade

Practice All PMLE Questions

Access 65 questions with complete answers and detailed explanations.

View Full PMLE Practice Test →

← Back to PMLE Study Guide