A genomics team runs an analysis that loads a 400 GB reference dataset entirely into memory and then makes many small passes over it. Processor usage stays moderate throughout and there is very little disk activity once the data is loaded. Which EC2 instance family should they start from?
AWS Certified Cloud Practitioner (CLF-C02), objective 3. Cloud technology and services medium
Machine-checked — no person has signed for it. This question was read against the source cited below by an automated pass, which found no contradiction. That is a weaker claim than it sounds: the same kind of process wrote the question, so it can confirm its own mistake.
Treat it as a good draft rather than as settled fact, and read the source below before you rely on it. It is not used in mock exams here — only questions a person has signed for are.
The options
Correct Memory optimized
Correct. The resource that runs out first is RAM, and memory optimized families give the most memory per instance and per vCPU. That is exactly what a large in-memory working set needs.
Not correct Compute optimized
Wrong. Compute optimized instances give a high ratio of vCPU to memory for processor-bound work. Here you would hit the memory ceiling long before you used the processors you were paying for.
Not correct Storage optimized
Wrong. Storage optimized families exist for very high local disk throughput and IOPS — large sequential scans, or databases reading from local NVMe. This workload touches the disk once and then works out of RAM.
Not correct Accelerated computing
Wrong on the evidence given. GPU and other accelerator instances suit massively parallel maths such as model training or rendering. Nothing here says the algorithm can use an accelerator, and these are the most expensive instances to guess with.
Why
Choose the instance family by the resource that becomes the bottleneck, then choose the size within that family. General purpose is balanced and the sensible default; compute optimized favours processor-heavy work; memory optimized favours large in-memory datasets, caches and big relational databases; storage optimized favours local disk throughput; accelerated computing adds GPUs or purpose-built chips. Getting the family right matters more than getting the size right, because the size can be changed later in a stop and start.
Where this comes from
- Cited
- AWS exam guide task statement 3.3
Practise this
Reading one question is not practice. The trainer will draw a set from objective 3 and space the ones you get wrong.
Practise AWS Certified Cloud Practitioner (CLF-C02)
More questions on this objective
- A team builds the same stack — a load balancer, three EC2 instances and a database — by clicking through the console once for dev, once for test and once for production. The three environments have quietly drifted apart and a bug that only appears in production took a week to trace to a missing setting. They want the stack described once, reviewed in a pull request, and applied identically to each environment. Which service does that? machine-checked
- A Python reporting job runs unattended every night. Inside its own logic it needs to list objects in an S3 bucket, read a few of them, and write a summary object back, handling failures with the program's existing retry and logging code. What is the appropriate way for that program to call AWS? machine-checked
- A small team has a Java web application and no operations staff. They want to hand AWS the application package and have it provision the instances, the load balancer and the scaling policy, while keeping the ability to inspect and adjust those resources later if they need to. Which service fits? machine-checked
- A manufacturer keeps its plant-floor control application on servers inside the factory, because production must continue when the internet link drops. Its reporting and analytics run in AWS and read the same production data. Which TWO statements about this arrangement are accurate? machine-checked
- An internal application runs on a single EC2 instance in one Availability Zone. During a power event in that Availability Zone the application was unreachable for four hours. The team wants the cheapest change that keeps it serving through the same kind of event, without moving to another Region. What should they do? machine-checked
- A German insurer is told by its regulator that customer records must remain within the European Union. Four candidate Regions can all run every service the workload needs, and one of them is noticeably cheaper per instance-hour. Which consideration settles the choice of Region first? machine-checked