SAP-C02 Design for New Solutions Practice Question

A financial services firm is migrating a latency-sensitive trading analytics workload to AWS. The workload runs on Amazon EC2 instances in a single VPC and requires network throughput of up to 100 Gbps per instance, consistent sub-100-microsecond inter-instance latency, and the ability to place instances close together to minimize hops. The firm also wants the instances to communicate without traversing the public internet. (Choose two.)

⚠ Common exam trap

The trap here is reaching for a global traffic-optimization service for inter-instance latency when the requirement is really about physical placement and the network interface technology inside a single VPC.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Attach an Elastic Fabric Adapter (EFA) to the instances and enable it for the application

Low-latency, high-bandwidth east-west traffic is achieved by combining a cluster placement group, which minimizes physical distance and hops within one Availability Zone, with Elastic Fabric Adapter, which provides OS-bypass networking for tightly coupled workloads. Together they deliver the required per-instance throughput and consistent latency while keeping traffic on the private AWS network.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Configure a VPC peering connection between the workload VPC and a shared services VPC in another Region
Why it's wrong here
VPC peering connects separate VPCs and, across Regions, routes traffic over the AWS backbone with additional latency. It does not improve throughput or latency between instances inside the same VPC, and it does not address the placement or enhanced networking requirements, so it fails to meet the scenario's performance goals.
✗
Deploy the instances into a spread placement group across three Availability Zones
Why it's wrong here
Spread placement groups deliberately separate instances across distinct hardware and Availability Zones to reduce correlated failure, which increases network distance and latency. That directly conflicts with the requirement for consistent sub-100-microsecond latency and close physical placement, so it is the wrong placement strategy for this workload.
✓
Attach an Elastic Fabric Adapter (EFA) to the instances and enable it for the application
Why this is correct
EFA provides a network interface that supports OS-bypass for tightly coupled HPC and analytics workloads, delivering lower and more consistent latency than standard TCP paths. It is designed for the high per-instance throughput the workload demands and works with cluster placement groups to keep traffic private and within a single Availability Zone.
✗
Enable AWS Global Accelerator with endpoint groups pointing to the instances in multiple Regions
Why it's wrong here
Global Accelerator optimizes ingress from global clients over the AWS backbone, but it does not lower inter-instance latency inside a VPC and does not provide the 100 Gbps east-west throughput the workload needs. Routing traffic through accelerator endpoints would also add hops rather than remove them, so it does not satisfy the placement or bandwidth requirements.
✓
Launch the instances into a cluster placement group within a single Availability Zone
Why this is correct
A cluster placement group packs instances close together inside one Availability Zone, which reduces network hops and delivers the low, consistent latency the trading analytics workload needs. Combined with high-bandwidth instance types that support enhanced networking, it enables the multi-gigabit throughput required while keeping traffic on the private AWS network rather than the public internet.

Go deeper

SAP-C02 Design for New Solutions Practice Question

Why each option matters

Option-by-option breakdown

Related to this question