PMLE Serving and Scaling Models Practice Question
A team wants to run a Vertex AI pipeline that deploys a model, runs a smoke test against the endpoint, and automatically rolls back if the smoke test fails. They need the deployment step to be reversible and the endpoint to remain available during the update. Which approach should they use?
⚠ Common exam trap
The trap here is treating redeployment after failure as equivalent to a rollback, when a true rollback leaves the prior deployment serving continuously with no gap.
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
✓
Deploy the new model as a second deployment resource on the endpoint with 0% traffic, run the smoke test, then shift traffic and delete the old deployment if the test passes.
Adding a new deployment resource to the existing endpoint lets the team validate the new model without disrupting current traffic. Traffic can be moved only after the smoke test succeeds, and removing the old resource completes the rollout; if the test fails, the old resource is still serving and no rollback action is needed.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Deploy the new model to a separate endpoint, run the smoke test there, and then update the application's endpoint URL after the test passes.
Why it's wrong here
A separate endpoint isolates the new model but requires clients to change URLs, which is disruptive and not an atomic traffic shift. It also leaves the original endpoint running as a separate resource that must be cleaned up. The scenario calls for a reversible deployment on the endpoint, not a client-side endpoint swap.
- ✗
Use a Vertex AI pipeline with a custom training component that trains and deploys the model in a single step, and rely on pipeline retry policies for rollback.
Why it's wrong here
Pipeline retries re-execute failed steps but do not provide a controlled traffic shift or a clean rollback of an endpoint deployment. Combining training and deployment in one component also makes the smoke test harder to isolate. This does not satisfy the reversible, always-available deployment requirement.
- ✓
Deploy the new model as a second deployment resource on the endpoint with 0% traffic, run the smoke test, then shift traffic and delete the old deployment if the test passes.
Why this is correct
Creating a new deployment resource on the same endpoint keeps the existing model serving while the new one is validated. Traffic can be shifted only after the smoke test passes, and the old resource can be removed afterward. This pattern preserves availability and makes rollback trivial because the old deployment is untouched until the new one is proven.
- ✗
Undeploy the current model, deploy the new model on the same endpoint, and run the smoke test; if it fails, redeploy the previous model.
Why it's wrong here
Undeploying first creates a window with no model serving, violating the availability requirement. Redeploying the previous model after a failure adds further downtime and may not restore the exact prior state quickly. This approach is simpler but directly conflicts with keeping the endpoint available during the update.
Go deeper
Related to this question
About these practice questions
This PMLE question is part of Courseiva's 775-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 and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Google Cloud exam blueprint
This PMLE 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 PMLE exam.