Home › Forums › AWS › AWS Certified Solutions Architect Professional › Incorrect answer on EBS Cold HDD
-
On this question:
Q: A company is migrating a legacy Oracle database from their on-premises data center to AWS. It will be deployed in an existing EBS-backed EC2 instance with multiple EBS volumes attached. For the migration, a new volume must be created for the Oracle database and then attached to the instance. This will be used by a financial web application and will primarily store historical financial data that are infrequently accessed.Which of the following is the MOST cost-effective and throughput-oriented solution that the solutions architect should implement?
The answer marked as correct was:
A: Migrate the database using the AWS Database Migration Service and use a Cold HDD (sc1) EBS volume.
There is no way someone would select Cold HDD for a solution that was described as “throughout-oriented”. I suspect the question write was fooled by the fact that the documentation for Cold HDD starts with a generic statement that says, “Cold HDD volumes … defines performance in terms of throughput rather than IOPS. “. This doesn’t say performance will be good. In fact, later in the documentation, AWS states, “making these volumes ideal for IO intensive workloads with relatively low baseline throughput requirements.“. So definitely NOT something one would call a “throughput-oriented solution”. I selected Provisioned IOPS, which I know is not ideal for infrequent access, but at least it doesn’t completely contradict a basic requirement.
-
Hello fernando-6
Thanks for your feedback.
First off, ‘throughput-oriented’ does not necessarily mean ‘high-performance’. It’s simply a description of how data is processed, focusing on the volume of data that can be handled over time, rather than the speed of each individual operation — something Cold HDD is also capable of. Yes, Cold HDD won’t provide the same level of performance as some of the other storage types, but this is the trade-off. It strikes a balance between affordability and performance, which is essential for data that isn’t accessed often. Our goal here isn’t to chase the highest possible performance regardless of cost; instead, we’re aiming for a practical option that can still meet the nature of the workload being described.
Let me know if this helps.
Regards,
Carlo @ Tutorials Dojo
-
No Carlos, thanks for responding, but that doesn’t help. In any reasonable interpretation of the English language, if someone says they want a solution that is “throughput-oriented”, clearly they want something that is NOT bad on throughput. Maybe doesn’t necessarily means it has to have the best throughput in the world, but it cannot be bad at it. As I pointed it out, the AWS documentation explicitly states Cold HDD would only work if you had a LOW throughput requirement.
As I said in my comment, I WASN’T chasing the best performance. In fact I said the answer I picked wasn’t ideal. But at least it doesn’t specifically contradict a requirement. If you want to make the question valid, you need to remove that “throughput-oriented” from it.
-
Generally I agree with the take that throughput oriented means getting better performance/throughput.
However, I think for this one specifically though in AWS cert context, if you look @ https://aws.amazon.com/ebs/cold-hdd/, the use case section has exactly this language: “Throughput-oriented storage for data that is infrequently accessed”.
I think the other hint on the question here that drives to this as the correct answer is the last statement that it is used to store historical data & infrequently accessed.
TLDR; I don’t like that language being used either, but it’s pulled direct from AWS doc page.
- This reply was modified 11 months, 1 week ago by CexamR.
- This reply was modified 11 months, 1 week ago by CexamR.
- This reply was modified 11 months, 1 week ago by CexamR.
- This reply was modified 11 months, 1 week ago by CexamR.
- This reply was modified 11 months, 1 week ago by CexamR.
- This reply was modified 11 months, 1 week ago by CexamR.
-
Yes, I mentioned in my original post that the expression had been copy/pasted from an article. But in the article, it was one expression in the context of a whole paragraph. And the whole paragraph explained that Cold HDD was only an option for low throughput requirements. That’s the danger of just copy/pasting terms without context.
In the exam prep question, the term was used in the last sentence, when the question summarizes the key requirements for the solution. So in that context, clearly it called for a solution that supports good throughput.
I’ll stop arguing. As I put in a separate post, this is one of many poorly worded or flat out wrong questions. I started posting them, but them I got tired because there were too many. I’ve purchased many TutorialDojo exam preps, and I always speak highly of them. This particular set is unfortunately below average. I’m still using as a study guide, because it helps me focus on the different areas of the test. But the quality in this one is subpar.
-
Hello fernando-6,
Thank you so much for your continued support in using our practice materials. We appreciate customers like you who are willing to speak up because it helps us understand and improve our products better.
I understand your points, and I’d like to clarify our stance. I believe that defining how ‘bad’ a throughput is should be relative to the presented requirements rather than taking it at face value. In context, while Cold HDD might not be ideal for high-performance scenarios, it may suffice for low-throughput needs. Regarding your claim of the poor usage of the term ‘throughput-oriented’; In discussing EBS volumes, we primarily focus on two performance-defining types: IOPS and throughput, each tailored for specific use cases. Our intention in using the term ‘throughput-oriented’ is to clearly indicate that we’re seeking a solution that prioritizes data transfer over large volumes. If someone asked me to pick a volume for reading hundreds of MBs per block, wouldn’t a throughput-oriented type be better than IOPS?
For this reason, I’m puzzled as to how Cold HDD would contradict the requirements of this scenario in any manner. As you’ve already said, Provisioned IOPS wouldn’t be ideal; it would be overkill for the job and more expensive. But what if there’s an alternative like Cold HDD that matches our workload’s demands and comes at a lower cost? Wouldn’t that be a more appropriate solution?
-
-
Carlo, I said I would stop responding, but I’ll make a final exception. You said you are willing to improve, but fighting this one the way you are, doesn’t really show that. This should be a clear cut, “Yeah, you’re right, we’ll improve the text”.
Your question ends with, “Which of the following is the MOST cost-effective and throughput-oriented solution that the solutions architect should implement?”. So the two main requirements are:
- Cost-effective is a subjective term. Generally it means you want something as cheap as possible, but “possible” depends on the requirements. If you’re doing an ML workload, using a T2-micro instance won’t work, regardless of how cheap that might be. So “cost-effective” generally means looking at the options available that fit my requirements, and pick the cheapest one among that.
- Throughput-oriented, in any normal interpretation, would mean a solution that focus on throughput. So throughput should be a high priority. Cold HDD is literally the lowest throughput option available (tied with gp2). How can anyone possibly say that a “throughput-oriented” solution, is one using the lowest throughput available?
I’m not telling you you folks need to toss the entire question. That are a couple simple tweaks that would make it correct. One, you could remove the “throuput-oriented” in the last line. Or if you want to keep it there, at least change the option from Cold HDD to Throughput Optimized HDD.
That’s it. I’m definitely done now.
-
Hi Fernando,
First off, thank you for bringing this up. I personally appreciate your time in discussing your thought process for this particular scenario, and I do understand your concern, as well as your decision not to reply to this discussion anymore.
I would also like to accentuate the response of CexamR and Carlo here. Whatever the outcome of our discussion here, the rationale that is backed up by the official AWS documentation should prevail.
Let me respond to these things that you said about the “throughout-oriented” and “infrequently accessed” keywords:
There is no way someone would select Cold HDD for a solution that was described as “throughout-oriented”.
One, you could remove the “throuput-oriented” in the last line. Or if you want to keep it there, at least change the option from Cold HDD to Throughput Optimized HDD.
It’s quite clear in the official documentation that Cold HDD is the MOST suitable solution for:
https://media.tutorialsdojo.com/throughput-oriented-storage.png
“Throughput-oriented storage for data that is infrequently accessed”
Those two keywords are there in verbatim. Moreover, the requirement of “MOST cost-effective” solution reinforces the need to use “Cold HDD”, just as mentioned in the table of this official AWS documentation: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ebs-volume-types.html#vol-type-hdd
-
Throughput-oriented storage for data that is infrequently accessed
-
Scenarios where the lowest storage cost is important”
Screenshot: https://media.tutorialsdojo.com/st1-vs-sc1.png
You kept on insisting that Throughput Optimized HDD volumes (st1) is the correct answer. But you haven’t provided a strong documentation for this.
The official AWS documentation mentions that a Throughput Optimized HDD volume is ideal for frequently accessed, throughput-intensive workloads.
Consider this – Does the scenario ever mentioned that…
– The data is “frequently accessed”?
– The workload is throughput-intensive?
Take note that throughput-oriented is different from throughput-intensive. You are pushing for a solution that provides more throughput, even though it is EXPLICITLY NOT required in the scenario.
We will let the readers decide whether they will believe the official AWS documentation or your unsubstantiated opinion.
Let us know if you need further assistance. The Tutorials Dojo team is dedicated to help you pass your AWS exam on your first try!
Regards,
Jon Bonso @ Tutorials Dojo
-
-
<div>Just my two cents…</div>
This scenario clearly mentions 3 points:
– Throughput-oriented workload
– Lowest Cost
– Infrequent Access
AFAIK, “Throughput” is the dominant performance attribute for HDDs while it is “IOPS” for SSDs so the first keyword can refer to either st1 (throughput optimized HDD) or sc1 (Cold HDD).
The other 2 keywords are clearly in favor for Cold HDD since it’s been mentioned almost all AWS docs that it is meant for infrequently access workloads and for cases where the lowest cost is preferred:
https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/hdd-vols.html
I got this question right the first time because of those keywords when I’m doing my review. The word “throughput” basically gives you the distinction in using HDD over SSD since SSD is more of “IOPS-oriented” rather than “throughput-oriented”.
I recommend to Fernando-6 to carefully consider the other related keywords and give more weight on the keywords at the end of the question which in this case, “MOST COST-EFFECTIVE”
Log in to reply.