Question 490
A team has corrected this HPE compute solution issue: A customer requests GPUs for all microservices even though most data-processing steps are CPU-bound. Which continuing practice would best prevent the same problem?
Explanation
Correct Answer
Size CPU, GPU and I/O stages as a coordinated AI pipeline. This recurring assurance measure is tied to the diagnosed risk: Misplaced acceleration can waste resources without improving pipeline throughput. For the active incident, the corrective action remains: Profile preprocessing, model execution and postprocessing separately before hardware allocation.
Incorrect Answers
Hardware utilization alone does not reveal application inference bottlenecks. The option addresses that risk, not the one responsible for the present issue. Misplaced acceleration can waste resources without improving pipeline throughput. Size CPU, GPU and I/O stages as a coordinated AI pipeline.
The prevention measure is relevant to this condition: Context length changes memory use, throughput and cost for a serving endpoint. Here the recurring weakness is: Misplaced acceleration can waste resources without improving pipeline throughput. Size CPU, GPU and I/O stages as a coordinated AI pipeline.
End-to-end RAG depends on storage, vector search and embedding compute as well as generation. The selected safeguard helps with that exposure, but it leaves the recurrence risk here unaddressed. Misplaced acceleration can waste resources without improving pipeline throughput. Size CPU, GPU and I/O stages as a coordinated AI pipeline.