Examkin

Soru

Zorluk: ZorHigh-Performing and Scalable Storage Solutions

A genomics research facility is migrating its high-performance computing (HPC) pipeline to AWS. The pipeline runs on a cluster of 3232 Amazon EC2 instances and requires concurrent, shared read and write access to a temporary scratch workspace. The storage system must deliver sub-millisecond latencies, support parallel I/O, and scale aggregate throughput linearly up to 15 GB/s15\text{ GB/s} during peak processing. Which two actions should a solutions architect take to meet these performance requirements at the lowest cost and operational complexity? (Select TWO.)

  1. Deploy an Amazon FSx for Lustre file system using the Scratch-2 deployment type.Cevap
  2. Mount the file system on the cluster nodes using the open-source Lustre client.Cevap
  3. C
    Deploy an Amazon EFS file system configured with Provisioned Throughput and mount it via NFSv4.
  4. D
    Deploy an Amazon EBS io2 Block Express volume with Multi-Attach enabled and format it with a standard ext4 file system.
  5. E
    Deploy an Amazon S3 bucket and mount it as a local file system using the Amazon S3 File Gateway on each EC2 instance.

Cevap

Deploy an Amazon FSx for Lustre file system using the Scratch-2 deployment type and mount the file system on the cluster nodes using the open-source Lustre client.
The correct architecture utilizes Amazon FSx for Lustre in the Scratch-2 deployment configuration along with the native Lustre client. Lustre is a parallel file system designed for HPC workloads, allowing multiple EC2 instances to read and write to the same storage blocks concurrently with sub-millisecond latencies. The Scratch-2 model is ideal for temporary processing data because it minimizes cost by omitting replication. Mounting with the native Lustre client ensures that parallel I/O is achieved across all 32 nodes.

Adım Adım Çözüm

1
Analyze the workload requirements: temporary scratch workspace, 32 concurrent nodes requiring shared read/write, sub-millisecond latencies, parallel I/O, and up to 15 GB/s15\text{ GB/s} throughput.
Identified the need for a high-performance, parallel, and temporary shared file system.
HPC clusters performing genomic processing require scalable, highly parallel file systems that can handle concurrent client access without data corruption or bottlenecks.
2
Select the correct AWS storage service that provides native parallel file system capabilities and optimized scratch deployment.
Selected Amazon FSx for Lustre with the Scratch-2 deployment type.
Lustre is designed for high-performance computing, and its Scratch-2 model offers a cost-effective, high-throughput solution without replication overhead, perfect for temporary scratch files.
3
Determine the optimal client mounting protocol to leverage the file system's parallel architecture.
Chose the open-source Lustre client for mounting the file system.
Standard NFS clients do not support the distributed, parallel architecture of Lustre, whereas the native Lustre client enables parallel data paths from all 32 EC2 instances to the storage servers.

Anahtar Kavram

Amazon FSx for Lustre is purpose-built for high-performance computing (HPC) workloads requiring high throughput, low latency, and parallel file access across many compute nodes. For temporary workloads, Scratch-2 deployment provides the best cost-to-performance ratio.
Bu soruyu puanla

Konular

High-Performing and Scalable Storage Solutions

Bu konudan daha fazla soru