An enterprise has consolidated billing enabled under AWS Organizations with 3 active AWS a… — AWS Certified Solutions Architect - Professional

An enterprise has consolidated billing enabled under AWS Organizations with $3$ active AWS accounts:

* Account A (Web Application): Hosts an API on Amazon ECS on AWS Fargate. The tasks run in private subnets across $3$ Availability Zones. The tasks download large application packages from Amazon S3 and pull container images from Amazon ECR. A single NAT Gateway is used in a public subnet to provide outbound internet access for the tasks to communicate with ECS, ECR, S3, and CloudWatch. No other outbound internet access is required.
* Account B (Data Processing): Runs a monthly batch-processing application on a fleet of Amazon EC2 `r6i.xlarge` instances ( $32\text{ GiB}$ RAM). CPU utilization averages $10\%$ and memory utilization averages $15\%$ for most of the month. However, during the $48\text{-hour}$ monthly run, memory utilization peaks at $82\%$ (approximately $26\text{ GiB}$ ) while CPU utilization reaches $45\%$ .
* Account C (Data Ingestion): Uses AWS Lambda functions to process streaming data and write it to Amazon DynamoDB.

The enterprise wants to minimize total AWS costs across all accounts. The architecture must remain highly available, support the performance requirements of the monthly batch processing without operational failures, and minimize operational overhead.

Which TWO actions should the Solutions Architect recommend to optimize resources and reduce costs? (Select TWO.)

In Account A, deploy a VPC Gateway Endpoint for Amazon S3 and VPC Interface Endpoints for Amazon ECS, Amazon ECR, and Amazon CloudWatch in the VPC. Configure the Fargate tasks to use these endpoints, and delete the NAT Gateway.Answer
Purchase a Compute Savings Plan at the AWS Organizations management account level. In Account B, implement an automated script to stop the EC2 instances, change their type to `r6i.large` ( $16\text{ GiB}$ RAM) during off-peak periods, and resize them back to `r6i.xlarge` ( $32\text{ GiB}$ RAM) only for the $48\text{-hour}$ monthly batch-processing window.Answer
C
Purchase an EC2 Instance Savings Plan at the AWS Organizations management account level to maximize the discount rate across the Amazon EC2, AWS Fargate, and AWS Lambda workloads in all member accounts.
D
In Account A, share the existing NAT Gateway with Account C by using AWS Resource Access Manager (RAM) to associate the NAT Gateway resource with Account C's VPC, thereby eliminating the hourly provisioning cost of a separate NAT Gateway in Account C.
E
To minimize outbound data transfer fees in Account A, configure the routing tables in all three Availability Zones to point to a single NAT Gateway located in a single public subnet, and purchase a Compute Savings Plan to cover the NAT Gateway hourly and data processing charges.

Answer

Deploy Gateway and Interface VPC Endpoints in Account A to replace the NAT Gateway, and purchase a Compute Savings Plan while dynamically resizing Account B's EC2 instances between `r6i.large` and `r6i.xlarge` to accommodate the monthly batch peak.

Replacing the NAT Gateway with VPC endpoints (Gateway for S3, Interface for ECS, ECR, and CloudWatch) eliminates the NAT Gateway hourly and data processing fees for Account A since the Fargate tasks do not require external internet access. Purchasing a Compute Savings Plan at the management account level applies discounts across Fargate, Lambda, and EC2 workloads in all accounts. In Account B, resizing the EC2 instances to `r6i.large` during off-peak times and resizing them to `r6i.xlarge` during the

48\text{-hour}

batch window ensures that the peak memory requirement of

26\text{ GiB}

is met without paying for overprovisioned resources for the rest of the month.

Step-by-Step Solution

Analyze Account A workloads and network costs.

Fargate tasks communicate only with AWS services (S3, ECR, ECS, CloudWatch). NAT Gateway costs can be eliminated by replacing them with a Gateway VPC Endpoint for S3 (which is free) and Interface VPC Endpoints (AWS PrivateLink) for ECS, ECR, and CloudWatch.

This eliminates NAT Gateway hourly charges and data processing fees ($0.045 per GB) which are typically high for container image pulls and package downloads.

Analyze Account B workloads and sizing.

The EC2 instances average low utilization (

10\%

CPU,

15\%

memory) but peak at

82\%

memory (approx.

26\text{ GiB}

) during a

48\text{-hour}

window. Resizing the instances to `r6i.large` (

16\text{ GiB}

) during off-peak times and resizing them to `r6i.xlarge` (

32\text{ GiB}

) during the batch run meets the requirements without causing out-of-memory failures.

If they were permanently resized to a smaller size like `r6i.large` or a burstable type like `t3.xlarge` (

16\text{ GiB}

), the batch job would fail due to insufficient memory during the peak run.

Determine the appropriate Savings Plan type and scope.

A Compute Savings Plan purchased at the management account level provides the highest flexibility, covering EC2, Fargate, and Lambda across all member accounts.

EC2 Instance Savings Plans are limited to EC2 and do not cover Fargate or Lambda workloads. Consolidating the purchase at the management account level allows the discount to apply to all accounts under the consolidated billing organization.

Key Concept

Multi-account cost optimization utilizing Compute Savings Plans, dynamic EC2 right-sizing based on peak memory demands, and replacing NAT Gateways with VPC Endpoints for AWS services.

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics

More questions from this topic