AWS Certified Solutions Architect – Professional — Question 654

A company asks a solution architect to optimize the cost of a solution. The solution handles requests from multiple customers. The solution includes a multi-tier architecture that uses Amazon API Gateway, AWS Lambda, AWS Fargate, Amazon Simple Queue Service (Amazon SQS), and Amazon EC2.
In the current setup, requests go through API Gateway to Lambda and either start a container in Fargate or push a message to an SQS queue. An EC2 Fleet provides EC2 instances that serve as workers for the SQS queue. The EC2 Fleet scales based on the number of items in the SQS queue.
Which combination of steps should the solutions architect recommend to reduce cost the MOST? (Choose three.)

Answer options

• A. Determine the minimum number of EC2 instances that are needed during a day. Reserve this number of instances in a 3-year plan with payment all upfront.
• B. Examine the last 6 months of compute utilization across the services. Use this information to determine the needed compute for the solution. Commit to a Savings Plan for this amount.
• C. Determine the average number of EC2 instances that are needed during a day. Reserve this number of instances in a 3-year plan with payment all upfront.
• D. Remove the SQS queue from the solution and from the solution infrastructure.
• E. Change the solution so that it runs as a container instead of on EC2 instances. Configure Lambda to start up the solution in Fargate by using environment variables to give the solution the message.
• F. Change the Lambda function so that it posts the message directly to the EC2 instances through an Application Load Balancer.

Correct answer: B, D, E

Explanation

Committing to an AWS Savings Plan (Option B) provides broad savings across multiple compute types including Lambda and Fargate, which is more effective than rigid EC2 Reserved Instances. Eliminating the SQS queue (Option D) and migrating the EC2 worker instances to AWS Fargate containers (Option E) allows the architecture to become fully serverless, removing the overhead and idle costs of managing EC2 fleets. This shift ensures the company only pays for compute resources when they are actively processing requests.