A company is using Amazon SageMaker and millions of files to train an ML model. Each file is several megabytes in size. The files are stored in an Amazon S3 bucket. The company needs to improve…

Question

Options

• ATransfer the data to a new S3 bucket that provides S3 Express One Zone storage. Adjust the
• BCreate an Amazon FSx for Lustre file system. Link the file system to the existing S3 bucket.
• CCreate an Amazon Elastic File System (Amazon EFS) file system. Transfer the existing data to
• DCreate an Amazon ElastiCache (Redis OSS) cluster. Link the Redis OSS cluster to the existing

• A
  8% (4)
• B
  81% (43)
• C
  9% (5)
• D
  2% (1)

Explanation

Option B is correct because Amazon FSx for Lustre is purpose-built for high-performance computing and ML workloads - it provides a high-throughput, low-latency parallel file system that integrates natively with S3. When linked to an S3 bucket, FSx for Lustre can ingest the data and serve it to SageMaker at speeds far exceeding standard S3 access, dramatically reducing training time with minimal setup effort.

Why the others fall short:

• A (S3 Express One Zone): While faster than standard S3, it still operates as object storage. Millions of small file reads over S3 (even Express) can't match the throughput of a parallel file system like Lustre for ML training workloads.
• C (Amazon EFS): EFS is a general-purpose NFS file system that works, but it doesn't offer the same high-throughput parallel I/O that Lustre is specifically optimized for - and migrating data to EFS takes more time than simply linking FSx to an existing bucket.
• D (ElastiCache Redis): Redis is an in-memory key-value store designed for caching application data and low-latency lookups, not for streaming millions of multi-megabyte training files to ML instances.

Memory tip: Think FSx for Lustre = ML/HPC speed demon. "Lustre" literally means Linux + cluster - whenever you see SageMaker + S3 + "need faster training I/O," FSx for Lustre is almost always the answer because it links to S3 with one step and acts as a high-speed cache layer.

No community discussion yet for this question.