Medium1 markMultiple Choice
Domain 2.5: PerformancePerformanceHPCNetworking

AWS SAP-C02 · Question 25 · Domain 2.5: Performance

A company runs a high-performance computing (HPC) workload on a cluster of Amazon EC2 instances. The instances need to communicate with each other with the absolute lowest possible latency and highest packet rate. The workload is tightly coupled and runs in a single Availability Zone. Which network configuration should the Architect use?

Answer options:

A.

Deploy the instances in a Cluster Placement Group and attach Elastic Fabric Adapters (EFA) to the instances.

B.

Deploy the instances in a Spread Placement Group and attach Elastic Network Adapters (ENA) to the instances.

C.

Deploy the instances across multiple Availability Zones and use AWS Global Accelerator for routing.

D.

Deploy the instances in a Cluster Placement Group and use VPC peering to route traffic between them.

How to approach this question

Combine the placement strategy for low latency (Cluster) with the network adapter designed for HPC (EFA).

Full Answer

A.Deploy the instances in a Cluster Placement Group and attach Elastic Fabric Adapters (EFA) to the instances.✓ Correct
Deploy the instances in a Cluster Placement Group and attach Elastic Fabric Adapters (EFA) to the instances.
For tightly coupled HPC workloads, you should use a Cluster Placement Group to ensure instances are physically located close to each other in the data center, minimizing network latency. Additionally, using an Elastic Fabric Adapter (EFA) provides OS-bypass capabilities, significantly reducing latency and increasing throughput compared to standard network interfaces.

Common mistakes

Confusing ENA (standard high performance) with EFA (HPC specific OS-bypass).

Practice the full AWS Solutions Architect Professional SAP-C02 Practice Exam 5

75 questions · hints · full answers · grading

More questions from this exam