New Year Sale 2026! Hurry Up, Grab the Special Discount - Save 25% - Ends In 00:00:00 Coupon code: SAVE25
Welcome to Pass4Success

- Free Preparation Discussions
Mail Us support@pass4success.com
Location US
Mob_nav_barsMENU

Amazon MLS-C01 Exam - Topic 1 Question 106 Discussion

Actual exam question for Amazon's MLS-C01 exam
Question #: 106
Topic #: 1
[All MLS-C01 Questions]

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?

Show Suggested Answer Hide Answer
Suggested Answer: C

The best solution for this scenario is to use shadow deployment, which is a technique that allows the company to run the new experimental model in parallel with the existing model, without exposing it to the end users. In shadow deployment, the company can route the same user requests to both models, but only return the responses from the existing model to the users.The responses from the new experimental model are logged and analyzed for quality and performance metrics, such as accuracy, latency, and resource consumption12. This way, the company can validate the new experimental model in a production environment, without affecting the current live traffic or user experience.

The other solutions are not suitable, because they have the following drawbacks:

A: A/B testing is a technique that involves splitting the user traffic between two or more models, and comparing their outcomes based on predefined metrics.However, this technique exposes the new experimental model to a portion of the end users, which might affect their experience if the model is not reliable or consistent with the existing model3.

B: Canary release is a technique that involves gradually rolling out the new experimental model to a small subset of users, and monitoring its performance and feedback.However, this technique also exposes the new experimental model to some end users, and requires careful selection and segmentation of the user groups4.

D: Blue/green deployment is a technique that involves switching the user traffic from the existing model (blue) to the new experimental model (green) at once, after testing and verifying the new model in a separate environment.However, this technique does not allow the company to validate the new experimental model in a production environment, and might cause service disruption or inconsistency if the new model is not compatible or stable5.

References:

1:Shadow Deployment: A Safe Way to Test in Production | LaunchDarkly Blog

2:Shadow Deployment: A Safe Way to Test in Production | LaunchDarkly Blog

3:A/B Testing for Machine Learning Models | AWS Machine Learning Blog

4:Canary Releases for Machine Learning Models | AWS Machine Learning Blog

5:Blue-Green Deployments for Machine Learning Models | AWS Machine Learning Blog


by Ira at Sep 16, 2024, 02:52 AM

Limited Time Offer

25%

Off

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

Contribute your Thoughts:

0/2000 characters
Submit Cancel
Glendora
3 months ago
Totally agree, shadow deployment keeps things safe!
upvoted 0 times
...
Glendora
3 months ago
Blue/green deployment sounds risky for a new model.
upvoted 0 times
...
Carissa
3 months ago
Wait, can you really validate without affecting live traffic?
upvoted 0 times
...
Glory
4 months ago
I think A/B testing could work too, but not ideal here.
upvoted 0 times
...
Felicidad
4 months ago
Shadow deployment is the way to go for this!
upvoted 0 times
...
Deangelo
4 months ago
Blue/green deployment sounds familiar, but I think it requires switching between two environments, which doesn't seem to align with the need to measure the new model without impacting users.
upvoted 0 times
...
Maryann
4 months ago
I practiced a similar question, and I believe canary release could be an option, but it usually involves gradually rolling out the new model, which might not fit the requirement of measuring without affecting live traffic.
upvoted 0 times
...
Rebecka
4 months ago
I'm not entirely sure, but I remember A/B testing being used for comparing models. However, it might not be suitable here since we can't affect current traffic.
upvoted 0 times
...
Kimberlie
5 months ago
I think shadow deployment might be the right choice since it allows the new model to run in the background without affecting live traffic.
upvoted 0 times
...
Beula
5 months ago
I'm pretty confident that the answer is B, canary release. This allows the company to gradually roll out the new model to a small subset of users, while the majority of traffic continues to be served by the existing model.
upvoted 0 times
...
Shaquana
5 months ago
Ah, I see. The question is specifically asking for a solution that meets the requirement of measuring the performance of the new model without affecting the current live traffic. I think the answer is C, shadow deployment.
upvoted 0 times
...
Clorinda
5 months ago
This seems like a classic deployment strategy question. I think the key is to identify the solution that allows the company to validate the new model without affecting the current live traffic.
upvoted 0 times
...
Bulah
5 months ago
Hmm, I'm a bit confused. The question mentions that only one model can serve user requests at a time, so I'm not sure if A/B testing would work here. Maybe a canary release or a shadow deployment would be a better fit?
upvoted 0 times
...
Elza
5 months ago
This looks like a pretty straightforward question. I think the key is to focus on the HPE OneView management aspect and look for an option that allows me to configure the workload profile across all the servers.
upvoted 0 times
...
Irene
1 year ago
As a model enthusiast, I'm all about option C. Shadow deployment is the perfect way to put the new model through its paces without causing any drama. Now, if only I could train my cat to be a machine learning expert...
upvoted 0 times
Salena
1 year ago
I've heard that option A is commonly used for testing different versions of software. It could be a good choice for this situation.
upvoted 0 times
...
Jolene
1 year ago
I think option B could also work well, gradually releasing the new model to a small subset of users.
upvoted 0 times
...
Rebbecca
1 year ago
I agree, option C is a great way to test the new model without disrupting the current system.
upvoted 0 times
...
...
Mari
1 year ago
Can you explain why you think Canary release is the best option?
upvoted 0 times
...
Glendora
1 year ago
This is a tough one, but I think option C is the winner. Shadow deployment sounds like the safest way to test the new model without causing any major disruptions. Kudos to the insurance company for being so cautious!
upvoted 0 times
...
Hobert
1 year ago
I'm feeling a bit mischievous today, so I'll go with option B - Canary release. Who doesn't love a little bit of controlled chaos, am I right?
upvoted 0 times
Mozelle
1 year ago
Definitely! It's a smart way to ensure the new model is ready to serve user requests without any disruptions.
upvoted 0 times
...
Sabra
1 year ago
I agree, it's a great way to validate the new model's performance in a production environment.
upvoted 0 times
...
Kanisha
1 year ago
Canary release sounds like a fun choice! It allows you to test the new model without affecting live traffic.
upvoted 0 times
...
...
Ashlyn
1 year ago
I disagree, I believe the correct answer is B) Canary release.
upvoted 0 times
...
Svetlana
1 year ago
Definitely, shadow deployment is the way to go here. Measure the performance of the new model without affecting the current users? Sign me up!
upvoted 0 times
Julianna
1 year ago
Canary release might be risky, shadow deployment is safer.
upvoted 0 times
...
Mozelle
1 year ago
A/B testing could also work, but shadow deployment is more suitable.
upvoted 0 times
...
Xochitl
1 year ago
I agree, it allows us to test the new model without impacting live traffic.
upvoted 0 times
...
James
1 year ago
Shadow deployment is the best option for this scenario.
upvoted 0 times
...
...
Mari
1 year ago
I think the answer is C) Shadow deployment.
upvoted 0 times
...
Anthony
1 year ago
Hmm, this seems like a classic case of trying to test the new model without disrupting the current live traffic. I'd go with option C - shadow deployment, seems like the most logical choice.
upvoted 0 times
Earleen
1 year ago
I agree, shadow deployment is the best solution for testing the new model in a production environment.
upvoted 0 times
...
Cheryl
1 year ago
Option C - shadow deployment allows the company to test the new model without affecting live traffic.
upvoted 0 times
...
...

Save Cancel