Amazon MLS-C01 Exam - Topic 4 Question 125 Discussion

Actual exam question for Amazon's MLS-C01 exam

Question #: 125
Topic #: 4

[Modeling]

An insurance company developed a new experimental machine learning (ML) model to replace an existing model that is in production. The company must validate the quality of predictions from the new experimental model in a production environment before the company uses the new experimental model to serve general user requests.

Which one model can serve user requests at a time. The company must measure the performance of the new experimental model without affecting the current live traffic

Which solution will meet these requirements?

AA/B testing

BCanary release

CShadow deployment

DBlue/green deployment

Show Suggested Answer

Suggested Answer: C

by Jacquline at Jan 05, 2026, 05:16 AM

Limited Time Offer

25%

Off

Get Premium MLS-C01 Questions as Interactive Web-Based Practice Test or PDF

Contribute your Thoughts:

Submit Cancel

Cordelia

4 days ago

Shadow deployment sounds interesting, but is it really effective?

upvoted 0 times

...

Skye

10 days ago

I think canary release is the best choice here.

upvoted 0 times

...

Niesha

15 days ago

A/B testing is a solid way to compare models!

upvoted 0 times

...

Maybelle

20 days ago

Definitely, they need to make sure they don't accidentally deploy the new model to the live environment and cause a disaster!

upvoted 0 times

...

Kattie

25 days ago

Haha, I bet the insurance company's IT team is having a field day with all these deployment options!

upvoted 0 times

...

Melvin

1 month ago

D) Blue/green deployment is more for switching between two complete environments, not just validating a new model.

upvoted 0 times

...

Valentin

1 month ago

A) A/B testing is not the right solution, as it requires serving user requests to both models simultaneously.

upvoted 0 times

...

Kenny

2 months ago

B) Canary release could also work, but it might be overkill for just validating the new model.

upvoted 0 times

...

Vince

2 months ago

I agree, C) Shadow deployment allows the company to measure the new model's performance without affecting the live traffic.

upvoted 0 times

...

Jaclyn

2 months ago

C) Shadow deployment seems like the best option here.

upvoted 0 times

...

Dorthy

2 months ago

Blue/green deployment could be a possibility too, but I feel like it might not fit the requirement of only serving one model at a time.

upvoted 0 times

...

Garry

3 months ago

Shadow deployment sounds familiar, but I can't recall if it really isolates the new model's performance from the live traffic.

upvoted 0 times

...

Quiana

3 months ago

I remember practicing with canary releases, and it seems like a good option since it allows gradual exposure of the new model.

upvoted 0 times

...

Sang

3 months ago

I think A/B testing might be the right choice, but I'm not entirely sure how it would work without affecting live traffic.

upvoted 0 times

...

Eladia

3 months ago

Hmm, I'm not sure. I think I need to re-read the question and the options more carefully. I want to make sure I understand the requirements and the differences between the deployment strategies before I commit to an answer.

upvoted 0 times

...

Titus

3 months ago

I'm leaning towards B) Canary release. That would allow us to gradually roll out the new model to a small subset of users, monitor its performance, and then expand it to more users if it's performing well. Seems like a safer approach than just switching over completely.

upvoted 0 times

...

Paris

3 months ago

Okay, I'm pretty confident that the answer is C) Shadow deployment. That way, we can run the new model in parallel with the existing one and measure its performance without affecting the live traffic. Seems like the perfect solution for this scenario.

upvoted 0 times

...

Daron

4 months ago

Hmm, I'm a bit confused. I know A/B testing is used to compare two versions, but I'm not sure how that would work in this case since we can only have one model in production at a time. I'll have to think this through more carefully.

upvoted 0 times

...

Catina

4 months ago

I think the key here is to find a way to test the new model without disrupting the current live traffic. Based on the options, I'd say Canary release or Shadow deployment seem like the best choices.

upvoted 0 times

...