Q58 — AWS SAA-C03 Ch.15

Question 58 of 100 | ← Chapter 15

Q1158. A machine learning (ML) team is building an application that uses data that is in an Amazon S3 bucket. The ML team needs a storage solution for its model training workflow on AWS.The ML team requires high-performance storage that supports frequent access to training datasets. The storage solution must integrate natively with Amazon S3.Which solution will meet these requirements with the LEAST operational overhead?

Correct Answer: C. Use Amazon FSx for Lustre to provide high-performance storage.Store training datasets in Amazon S3 Standard storage.

Explanation

To meet the requirements of high-performance storage for model training workflows that integrates natively with Amazon S3 and has the least operational overhead, the best solution is:Key Requirements:High-performance storage for frequent access to training datasets. Native integration with Amazon S3 (no manual data migration needed). Least operational overhead (minimal setup and management).Option Analysis:A. Use Amazon Elastic Block Store (Amazon EBS) volumes to provide high-performance storage. Use AWS DataSync to migrate data from the S3 bucket to EBS volumes. EBS is high-performance but not natively integrated with S3 (data must be copied manually or via tools like DataSync).AWS DataSync adds operational overhead (scheduling, monitoring data transfers).Not ideal for native S3 integration.B. Use Amazon EC2 ML instances to provide high-performance storage. Store training data on Amazon EBS volumes. Use the S3 Copy API to copy data from the S3 bucket to EBS volumes. EBS is high-performance but again requires manual data copying (using the S3 Copy API or CLI).Operational overhead (scripting, managing data copies).Not natively integrated with S3.C. Use Amazon FSx for Lustre to provide high-performance storage. Store training datasets in Amazon S3 Standard storage.Amazon FSx for Lustre is a high-performance, scalable file system optimized for ML workloads. It integrates natively with S3 (datasets can be referenced directly from S3 without manual copying). Least operational overhead (no data migration needed, just mount the FSx filesystem).Best fit for the requirements.D. Use Amazon EMR to provide high-performance storage. Store training datasets in Amazon S3 Glacier Instant Retrieval storage.Amazon EMR is for big data processing (e.g., Spark, Hadoop) but is not a storage solution for ML training. S3 Glacier Instant Retrieval is for archival storage (not high-performance, frequent access).Not suitable for ML training workflows.Correct Answer:C. Use Amazon FSx for Lustre to provide high-performance storage. Store training datasets in Amazon S3 Standard storage.This solution provides:High-performance storage (FSx for Lustre is optimized for ML). Native S3 integration (datasets can be referenced directly from S3).Least operational overhead (no data migration needed).Final Answer:C. Use Amazon FSx for Lustre to provide high-performance storage. Store training datasets in Amazon S3 Standard storage.The solution that best meets the requirements for high-performance storage that integrates natively with Amazon S3 and has the least operational overhead is:C. Use Amazon FSx for Lustre to provide high-performance storage. Store training datasets in Amazon S3 Standard storage.Amazon FSx for Lustre: This service provides a high-performance file system optimized for workloads like machine learning, particularly for applications that require fast access to large datasets. Integration with Amazon S3: FSx for Lustre can natively integrate with S3, allowing you to link your training datasets stored in S3 directly to the FSx file system, enabling seamless access and performance. Operational Overhead: Using FSx for Lustre simplifies data management, as it automatically syncs with S3, reducing the need for manual data migration or management tasks.This solution effectively supports the ML team's requirements for performance and ease of use.