Learn how to design an AWS architecture for HPC workloads requiring 10Gbps flows and 100Gbps aggregate throughput using Elastic Fabric Adapter and cluster placement groups for optimal performance.
Table of Contents
Question
A network engineer needs to design the architecture for a high performance computing (HPC) workload. Amazon EC2 instances will require 10 Gbps flows and an aggregate throughput of up to 100 Gbps across many instances with low-latency communication.
Which architecture solution will optimize this workload?
A. Place nodes in a single subnet of a VPC. Configure a cluster placement group. Ensure that the latest Elastic Fabric Adapter (EFA) drivers are installed on the EC2 instances with a supported operating system.
B. Place nodes in multiple subnets in a single VPC. Configure a spread placement group. Ensure that the EC2 instances support Elastic Network Adapters (ENAs) and that the drivers are updated on each instance operating system.
C. Place nodes in multiple VPCs Use AWS Transit Gateway to route traffic between the VPCs. Ensure that the latest Elastic Fabric Adapter (EFA) drivers are installed on the EC2 instances with a supported operating system.
D. Place nodes in multiple subnets in multiple Availability Zones. Configure a cluster placement group. Ensure that the EC2 instances support Elastic Network Adapters (ENAs) and that the drivers are updated on each instance operating system.
Answer
A. Place nodes in a single subnet of a VPC. Configure a cluster placement group. Ensure that the latest Elastic Fabric Adapter (EFA) drivers are installed on the EC2 instances with a supported operating system.
Explanation
This option leverages Elastic Fabric Adapter (EFA), which provides ultra-low latency and high bandwidth (up to 100 Gbps) networking for HPC workloads. Placing nodes in a cluster placement group and single subnet ensures optimal network performance as the instances are located close together with dedicated networking.
While other options could work, they do not maximize performance:
- B, D – Multiple subnets/AZs increase latency vs single subnet
- C – Transit Gateway adds networking hop vs single VPC
Therefore, solution A uniquely focuses the architecture specifically on ultra-low latency and high throughput needs of the HPC workload through EFA and cluster placement group.
AWS Certified Advanced Networking – Specialty ANS-C01 certification exam practice question and answer (Q&A) dump with detail explanation and reference available free, helpful to pass the AWS Certified Advanced Networking – Specialty ANS-C01 exam and earn AWS Certified Advanced Networking – Specialty ANS-C01 certification.