Examkin

Question

Difficulty: MediumHigh-Performing and Scalable Storage Solutions

A research organization is deploying a high-performance computing (HPC) cluster on AWS to process large-scale seismic simulation workloads. The cluster consists of multiple Amazon EC2 instances that require simultaneous, shared access to a storage volume. The storage solution must support parallel read/write operations, scale throughput up to 12 GB/s12\text{ GB/s}, and deliver sub-millisecond latencies.

Which combination of storage solutions and client configurations should a solutions architect select to meet these requirements? (Choose two.)

  1. Provision an Amazon FSx for Lustre file system with SSD storage using a persistent deployment model.Answer
  2. Mount the shared file system on the EC2 cluster using the native Lustre client to leverage parallel I/O.Answer
  3. C
    Provision an Amazon EFS file system configured with Elastic throughput and mount it using the Network File System (NFSv4) client.
  4. D
    Provision an Amazon EBS gp3 volume with maximum provisioned IOPS and throughput, and enable Multi-Attach to share the volume across the EC2 instances.
  5. E
    Create an Amazon S3 bucket and configure an S3 Gateway VPC Endpoint to mount it as a shared block device.

Answer

The correct answers are provisioning an Amazon FSx for Lustre file system with SSD storage and mounting the file system on the EC2 instances using the native Lustre client.
Amazon FSx for Lustre with SSD storage provides a high-performance parallel file system that is optimized for compute-heavy workloads like HPC, delivering sub-millisecond latencies and scaling to hundreds of gigabytes per second of throughput. Using the native Lustre client allows the EC2 instances to communicate directly and in parallel with the file system storage servers, maximizing throughput.

Step-by-Step Solution

1
Analyze the workload requirements: high-performance computing (HPC) seismic simulation cluster requiring shared storage access, parallel read/write, 12 GB/s12\text{ GB/s} throughput, and sub-millisecond latencies.
Identify that the workload demands an HPC-optimized parallel file system rather than a general-purpose file or object storage.
General-purpose solutions cannot meet the sub-millisecond latency and high parallel throughput requirements of this simulation cluster.
2
Select the appropriate storage service.
Choose Amazon FSx for Lustre with SSD storage using a persistent deployment model.
FSx for Lustre is purpose-built for HPC, offering sub-millisecond latency, massive parallel throughput, and high IOPS.
3
Determine the client mounting mechanism.
Choose to mount the file system using the native Lustre client.
The native Lustre client is required to exploit the parallel architecture of the Lustre file system to achieve the high throughput target.

Key Concept

High-Performance storage solutions for HPC workloads using Amazon FSx for Lustre and parallel clients
Rate this question

Topics

High-Performing and Scalable Storage Solutions

More questions from this topic