Your client is building a custom machine learning pipeline to identify lesions in the lungs based on x-rays. Different teams of data scientists are sharing common source data and building many ver…

Question

Options

• ARetain the data in use in a single region bucket with nearline storage. Retain the data in use in a
• BRetain the data in use in a single region bucket with standard storage.
• CRetain the data in use in a multi-region bucket.
• DRetain the data in use in a dual-region bucket.

• A
  9% (5)
• B
  83% (44)
• C
  6% (3)
• D
  2% (1)

Explanation

A single-region bucket with Standard storage is the best fit for an active ML pipeline where multiple teams are frequently reading and writing large datasets (X-ray images). Standard storage is designed for data that is accessed frequently and has no retrieval fees, making it cost-effective for high-throughput workloads. Nearline storage (Option A) is for data accessed less than once a month and incurs retrieval costs, which would be expensive for a busy ML pipeline. Multi-region (Option C) and dual-region (Option D) buckets add geo-redundancy and higher cost, which is not required here - co-located teams benefit more from the lowest latency of a single-region bucket.

No community discussion yet for this question.