Soru

Zorluk: ZorOptimizing Resource Costs and Sizing

An enterprise manages a multi-account environment within an organization in AWS Organizations. The consolidated billing feature is enabled.

- Account A runs a containerized batch-processing application on Amazon ECS with AWS Fargate, alongside short-lived webhook handlers running on AWS Lambda.
- Account B runs a set of high-availability backend APIs on Amazon EC2 m6i.xlarge\text{m6i.xlarge} instances (44 vCPUs, 16 GB16\text{ GB} memory) deployed across three Availability Zones. Average CPU utilization is consistently under 25%25\%, and memory utilization is under 30%30\%. The instances require at least 50 GB50\text{ GB} of local NVMe SSD storage for caching.
- Outbound internet traffic for both accounts currently flows through three NAT Gateways (one per Availability Zone) in each VPC.

The enterprise wants to optimize resource costs across both accounts while ensuring high availability for all workloads and outbound traffic paths with minimal operational effort.

Which TWO recommendations should the solutions architect make to achieve these goals? (Select TWO.)

  1. In Account B, change the EC2 instance type from m6i.xlarge to m6id.large to match the actual CPU and memory utilization while retaining the required local NVMe SSD storage.Cevap
  2. Purchase Compute Savings Plans in the management account of the organization to cover Fargate, Lambda, and EC2 compute costs across both member accounts.Cevap
  3. C
    Purchase EC2 Instance Savings Plans in the management account of the organization for the m6i instance family to maximize discounts across the EC2 instances in Account B and the Fargate tasks in Account A.
  4. D
    In both VPCs, consolidate outbound internet routing to use a single NAT Gateway in one Availability Zone to eliminate the hourly fees of the other NAT Gateways.
  5. E
    To reduce costs for partner-hosted databases, use AWS Resource Access Manager (RAM) to share Account B's subnets with the partner's external AWS account without enabling sharing with external entities in AWS Organizations.

Cevap

The correct recommendations are to right-size the EC2 instances in Account B to m6id.large and purchase Compute Savings Plans in the organization's management account.
Right-sizing the EC2 instances in Account B from m6i.xlarge to m6id.large successfully aligns the instance capacity with the actual resource consumption (which requires less than 1 vCPU and 4.8 GB memory) while retaining the critical local NVMe SSD storage requirement. Additionally, purchasing Compute Savings Plans in the organization's management account provides the necessary flexibility to automatically apply discounts across Fargate, Lambda, and EC2 instances across both member accounts.

Adım Adım Çözüm

1
Analyze resource utilization and constraints in Account B.
Peak utilization is under 1 vCPU1\text{ vCPU} (25%25\% of 44 vCPUs) and under 4.8 GB4.8\text{ GB} of RAM (30%30\% of 16 GB16\text{ GB}). A storage capacity of at least 50 GB50\text{ GB} of local NVMe SSD is required.
This establishes that the instances can be safely downsized to a smaller instance type within the same family that includes local NVMe storage (represented by the 'd' suffix).
2
Select the correct target instance type.
The m6id.large instance provides 22 vCPUs, 8 GB8\text{ GB} of RAM, and a 59 GB59\text{ GB} NVMe SSD.
It matches the resource limits and NVMe storage requirements without changing processor architectures, keeping operational effort low.
3
Determine the optimal Savings Plan type for a multi-account environment containing Fargate, Lambda, and EC2.
Compute Savings Plans are selected.
Compute Savings Plans apply globally across all accounts and cover EC2, Fargate, and Lambda, whereas EC2 Instance Savings Plans are limited to EC2 instances.
4
Evaluate and reject options that violate high availability or sharing constraints.
Reject NAT Gateway consolidation and invalid RAM configurations.
Reducing NAT Gateways creates a single point of failure, violating the high availability constraint. Subnet sharing to external accounts via RAM will fail without explicitly enabling sharing outside the organization.

Anahtar Kavram

Optimizing multi-account compute spend using Compute Savings Plans to cover mixed compute architectures (EC2, Fargate, Lambda) while right-sizing EC2 instances using historical metrics without losing critical hardware capabilities like local NVMe SSD storage.
Bu soruyu puanla