A network engineer needs to design the architecture for a high performance computing (HPC) workload. Amazon EC2 instances will require 10 Gbps flows…

The question requires a solution for a High-Performance Computing (HPC) workload with high-throughput (100 Gbps) and low-latency requirements. The optimal architecture for this is a cluster placement group, which co-locates instances on the same rack within a single Availability Zone to provide the lowest possible inter-node latency. The Elastic Fabric Adapter (EFA) is a network interface specifically designed for HPC and machine learning. It uses OS-bypass and the Scalable Reliable Datagram (SRD) protocol to deliver lower, more consistent latency and higher throughput than standard TCP/IP communication over an Elastic Network Adapter (ENA), making it essential for tightly-coupled HPC applications.

B: A spread placement group is designed for high availability by placing instances on distinct hardware, which increases latency and is counterproductive for HPC.

C: Routing traffic between VPCs via AWS Transit Gateway introduces significant network latency, making it unsuitable for tightly-coupled, low-latency HPC workloads.

D: A cluster placement group cannot span multiple Availability Zones. Its purpose is to group instances physically close together, which is only possible within a single AZ.

1. AWS Documentation, Amazon EC2 User Guide for Linux Instances, "Placement groups": "A cluster placement group is a logical grouping of instances within a single Availability Zone. ... This strategy enables workloads to achieve the low-latency network performance necessary for tightly-coupled node-to-node communication that is typical of HPC applications." This supports the use of a cluster placement group in a single AZ.

2. AWS Documentation, Amazon EC2 User Guide for Linux Instances, "Elastic Fabric Adapter": "An Elastic Fabric Adapter (EFA) is a network device that you can attach to your Amazon EC2 instance to accelerate High Performance Computing (HPC) and machine learning applications. EFA enables you to achieve the application performance of an on-premises HPC cluster, with the scalability, flexibility, and elasticity of the AWS Cloud." This confirms EFA is the correct network adapter.

3. AWS Documentation, Amazon EC2 User Guide for Linux Instances, "EFA basics": "To take full advantage of the EFA capabilities for tightly-coupled applications, launch all of your EFA-enabled instances in the same cluster placement group." This reference explicitly recommends the combination of EFA and cluster placement groups, as described in the correct answer.