AWS Certified Solutions Architect – Professional — Question 603
A medical company is running an application in the AWS Cloud. The application simulates the effect of medical drugs in development.
The application consists of two parts: configuration and simulation. The configuration part runs in AWS Fargate containers in an Amazon Elastic Container Service
(Amazon ECS) cluster. The simulation part runs on large, compute optimized Amazon EC2 instances. Simulations can restart if they are interrupted.
The configuration part runs 24 hours a day with a steady load. The simulation part runs only for a few hours each night with a variable load. The company stores simulation results in Amazon S3, and researchers use the results for 30 days. The company must store simulations for 10 years and must be able to retrieve the simulations within 5 hours.
Which solution meets these requirements MOST cost-effectively?
Answer options
- A. Purchase an EC2 Instance Savings Plan to cover the usage for the configuration part. Run the simulation part by using EC2 Spot Instances. Create an S3 Lifecycle policy to transition objects that are older than 30 days to S3 Intelligent-Tiering.
- B. Purchase an EC2 Instance Savings Plan to cover the usage for the configuration part and the simulation part. Create an S3 Lifecycle policy to transition objects that are older than 30 days to S3 Glacier.
- C. Purchase Compute Savings Plans to cover the usage for the configuration part. Run the simulation part by using EC2 Spot Instances. Create an S3 Lifecycle policy to transition objects that are older than 30 days to S3 Glacier.
- D. Purchase Compute Savings Plans to cover the usage for the configuration part. Purchase EC2 Reserved Instances for the simulation part. Create an S3 Lifecycle policy to transition objects that are older than 30 days to S3 Glacier Deep Archive.
Correct answer: C
Explanation
Compute Savings Plans are required for the configuration part because EC2 Instance Savings Plans do not cover AWS Fargate. EC2 Spot Instances are the most cost-effective choice for the simulation runner since the workload is short-lived and can handle interruptions. Finally, S3 Glacier meets the 5-hour retrieval window requirement (standard retrieval takes 3-5 hours), whereas S3 Glacier Deep Archive's standard retrieval takes up to 12 hours.