Question 520
A team has corrected this HPE compute solution issue: An HPE proposal uses a generic all-GPU configuration although the customer's main workload is CPU preprocessing and intermittent inference. Which continuing practice would best prevent the same problem?
Explanation
Correct Answer
Base final configuration on measured workload characteristics. This recurring assurance measure is tied to the diagnosed risk: The configuration overallocates acceleration without profiling the actual demand. For the active incident, the corrective action remains: Compare suitable validated configurations using workload-specific CPU and GPU measurements.
Incorrect Answers
The deployment statement lacks a responsible owner for site networking changes. The option addresses that risk, not the one responsible for the present issue. The configuration overallocates acceleration without profiling the actual demand. Base final configuration on measured workload characteristics.
The prevention measure is relevant to this condition: Infrastructure support and customer model-validation duties are different services. Here the recurring weakness is: The configuration overallocates acceleration without profiling the actual demand. Base final configuration on measured workload characteristics.
The configuration overallocates acceleration without profiling the actual demand. Using the proposed safeguard would guard against a separate risk: The attached service package lacks required backup and recovery planning expertise. Base final configuration on measured workload characteristics.