The auto-scaling would increase the rate at which sales requests are "processed", whereas a SQS will ensure messages don't get lost. If you were at a fast food restaurant with a long line with 3 cash registers, would you want more cash registers or longer ropes to handle longer lines? Same concept here.

"ensure that all the requests are processed successfully?" we want to ensure success not the speed, even in the auto-scaling, there is the chance for the failure of the request but not in SQS- if it is failed in sqs it is sent back to the queue again and new consumer will pick the request.

B doesn't fit because Auto Scaling alone does not guarantee that all requests will be processed successfully, which the question clearly asks for. D ensures that all messages are processed.

Hell true: I'd rather combine the both options: a SQS + auto-scaled bound to the length of the queue.

There are two components that we need * Frontend: Hosted on S3, performance can be increased with CloudFront * Backend: There's no reason to process all the orders instantly, so we should decouple the processing from the API which we do with SQS Thus D, CloudFront + SQS

I picked B before I read D option. Read the question again, it concerns:asynchronous processing of sales requests, Option D seems to align more closely with the requirements. So the requirement is ensuring all requests are processed successfully which means no request would be missed. So D is better option

I chose D because I love SQS! These questions are hammering SQS in every solution as a "protagonist" that saves the day. AC are clearly useless B can work but D is better because of SQS being better than EC2 scaling. The other part is that backend workers process the request asynchronously therefore a queue is better.

Amazon SQS will make sure that the requests are stored and didn't get lost. After that the workers asynchronously will process the requests. I would go for D

A and C don't solve anything so ignore them. Between B and D, D guarantees the scaling via SQS and order processing. B can also do that but it is not guaranteed that EC2 scaling will work to process the order. As usual, I suspect that this "brain dump" may be missing critical wording to differentiate between the options so read carefully in the exam.

And as others said, B might speed up the processing or reduce the number of lost orders, but we need to make sure that "ALL requests are processed successfully", NOT that "less requests are lost".

The problem states that the application consists of "static and dynamic front-end content." Static content typically includes cacheable resources such as HTML, CSS, and image files. Therefore, from this statement, one can infer that caching static content using CloudFront would improve performance. In other words, the mention of "static content" in the problem itself leads to the conclusion that CloudFront should be added for static content. Additionally, the problem mentions "asynchronously processed backend workers." Asynchronous processing is well-suited for services like SQS, which can improve efficiency by handling dynamic requests that do not require immediate processing. The mention of "successfully processing all requests" also suggests that SQS is needed to ensure that all requests are handled properly. Therefore, the correct answer is D.

An SQS queue acts as a buffer between the frontend (website) and backend (API). Web requests can dump messages into the queue at a high throughput, then the queue handles delivering those messages to the API at a controlled rate that it can sustain. This prevents the API from being overwhelmed.

D is correct.

Technically both option B and D would work. But, there's a need to process requests asynchronously, hence decoupling, hence Amazon SQS. I will settle with option D.

As an architecture, it is not possible to add more backend workers (it is part of the HR and boss's job, not for architecture design the solution). So when the demand surge, the only correct choice is to buffer them using SQS so that workers can take their time to process it successfully

D is correct.

eeeee... In the restaurant also here, the request in the line will be dropped in case of high-sudden volume flooding in so it will bring risk of missing request. while with SQS, it can ensure no lost but still slow processing with only 3 processor though so it is based on what is the question's concern. So I still voted for D.

D makes more sense

D make sens

Important question to answer D. Can you connect the website with SQS directly? How do you control access to who can put messages to SQS? I have never seen such a situation it has to be at least behind API gateway. So that conclusion brings me to answer B, application also can process async everything without SQS.

There is no clarity on what the asynchronous process is but D makes more sense if we want to process all requests successfully. The way the question is worded it looks like the msgs->SQS>ELB/Ec2. This ensures that the messages are processed but may be delayed as the load increases.

Options A and B would help by scaling out more instances, however, this may not scale quickly enough and still risks overwhelming the API. Caching parts of the dynamic content (option C) may help but does not provide the buffering mechanism that a queue does.

There is no scaling based on network traffic. since the backend worker works asynchronously SQS should be right here