Examkin

Question

Difficulty: Very hardHigh-Performing and Scalable Storage Solutions

An engineering team is deploying a parallel financial simulation workload on a fleet of clustered Amazon EC2 instances in a single Availability Zone. The application requires a POSIX-compliant shared file system capable of delivering sub-millisecond latencies, at least 85,00085,000 random IOPS, and 2 GB/s2\text{ GB/s} of throughput. The file system must dynamically load initial input datasets from an Amazon S3 bucket at startup and automatically write output results back to the same S3 bucket. Which storage configuration meets these requirements with the lowest operational complexity?

  1. An Amazon FSx for Lustre persistent file system integrated with the Amazon S3 bucket using a Data Repository Association (DRA).Answer
  2. B
    An Amazon Elastic File System (Amazon EFS) file system configured in Max I/O performance mode with Elastic throughput enabled.
  3. C
    An Amazon EBS Multi-Attach cluster of io2 Block Express volumes formatted with an ext4 file system across the EC2 fleet.
  4. D
    An Amazon FSx for OpenZFS file system configured with multiple NFS user shares and AWS DataSync for continuous S3 replication.

Answer

An Amazon FSx for Lustre persistent file system integrated with the Amazon S3 bucket using a Data Repository Association (DRA).
The configuration utilizing Amazon FSx for Lustre with a Data Repository Association (DRA) is correct because FSx for Lustre is optimized for compute-heavy, parallel workloads. It provides the required sub-millisecond latencies, massive scale-out IOPS and throughput, and natively handles bidirectional data synchronization with Amazon S3 automatically, which minimizes operational complexity.

Step-by-Step Solution

1
Evaluate the latency and throughput requirements of the workload (85,00085,000 IOPS, 2 GB/s2\text{ GB/s} throughput, sub-millisecond latencies).
Identify that general-purpose shared storage like Amazon EFS does not meet the sub-millisecond latency profile for intensive random IOPS workloads.
This narrows down the choices to high-performance block storage (EBS io2) or dedicated parallel file systems (FSx for Lustre).
2
Assess the shared access and file system requirements (POSIX-compliant, shared across multiple instances).
Recognize that utilizing Amazon EBS Multi-Attach with standard file systems (such as ext4) is invalid because standard file systems lack the locking mechanism needed to prevent data corruption.
This rules out the EBS option unless a complex cluster-aware file system (like GFS2) is deployed and configured, which contradicts the lowest operational complexity constraint.
3
Examine the integration with Amazon S3 for dynamic data loading and writing back.
Determine that Amazon FSx for Lustre natively supports Data Repository Associations (DRA) to bi-directionally sync with S3, while FSx for OpenZFS or EBS would require manual scripts or AWS DataSync configurations.
FSx for Lustre satisfies all performance, shared access, and S3 integration requirements with the lowest operational overhead.

Key Concept

High-Performance Shared Storage with S3 Integration
Estimated Time:3m 0s
Rate this question

Topics

High-Performing and Scalable Storage Solutions

More questions from this topic