mediumMultiple Select
PDE Practice Question: Which TWO configurations are required to enable…
Which TWO configurations are required to enable online prediction for a model deployed on Vertex AI Endpoints?
⚠ Common exam trap
A common mix-up: candidates confuse optional features (like Feature Store or autoscaling) with mandatory configurations, or assume the model must be trained on Vertex AI, when in fact only the machine type and model deployment are strictly required for online prediction.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
The endpoint must be configured with a machine type (e.g., n1-standard-2).
Option B is correct because a Vertex AI Endpoint requires a compute resource configuration, and deploying a model to an endpoint requires specifying a machine type (such as n1-standard-2) so the endpoint has the infrastructure to serve online prediction requests. Option D is correct because online prediction is only possible after a model is deployed to an endpoint; the DeployedModel resource on the endpoint is what actually receives and serves prediction traffic. Option A is incorrect because a Feature Store is not required for online prediction; it is only relevant when the model needs to retrieve features at serving time, and it is not a mandatory endpoint configuration. Option C is incorrect because a model can be imported and deployed to Vertex AI Endpoints even if it was trained outside Vertex AI, so training on Vertex AI is not required. Option E is incorrect because autoscaling is optional; an endpoint can serve online predictions with a fixed number of replicas without enabling autoscaling.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
A feature store must be attached to the endpoint.
Why it's wrong here
A feature store supplies training and serving features to models that consume them; it is not a prerequisite for deploying any model to an Endpoint. Feature stores are tempting because online serving often needs low-latency features, but the required configurations are the model's container image and the endpoint's deployed model resource.
- ✓
The endpoint must be configured with a machine type (e.g., n1-standard-2).
Why this is correct
Online prediction requires each deployed model to have compute resources, so the endpoint's deployed model must specify a machine type such as n1-standard-2. Without an assigned machine type, Vertex AI cannot provision serving nodes, and the endpoint cannot return synchronous predictions.
- ✗
The model must be trained on Vertex AI.
Why it's wrong here
Training location is irrelevant to serving: a model trained elsewhere can still be uploaded and deployed to an Endpoint for online prediction. Vertex AI training is tempting because it simplifies artefact registration, but the required configurations concern the deployed model's container and the endpoint's prediction serving, not its origin.
- ✓
A model must be deployed to an endpoint.
Why this is correct
Online prediction is served through an endpoint, so a model must first be uploaded to the Vertex AI Model Registry and then deployed to that endpoint. Deployment creates the serving infrastructure; without it, no prediction requests can be routed.
- ✗
Autoscaling must be enabled.
Why it's wrong here
Autoscaling adjusts replica count in response to traffic; a single-replica endpoint still serves online predictions. It is tempting because production endpoints usually enable it, but the required configurations are the model artefact with its serving container and the endpoint's deployed model, not scaling policy.
Go deeper
Related to this question
About these practice questions
This PDE question is part of Courseiva's 747-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This PDE practice question is part of Courseiva's free Google Cloud certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the PDE exam.