QuestionQ8

Overview of Huawei's AI Development Strategy and Full-Stack, All-Scenario AI Portfolio

Assume that a government customer has purchased two Huawei all-flash storage devices for service deployment. While using the storage devices, an engineer discovers that storage performance fluctuates frequently. To troubleshoot the storage-performance fluctuation issue, the engineer must perform many operations.

  1. Check whether there is an obvious change in the I/O model or service load when performance fluctuates.
  2. Use FTDS, the storage-side latency-location tool, to analyze the modules with fluctuating latency.
  3. Check whether the hit rate changes abruptly when performance fluctuates.
  4. Check whether extremely small I/Os are present when performance fluctuates.
  5. Check whether there are obvious hardware-resource bottlenecks on the storage side when performance fluctuates.
  6. Check whether there are obvious storage background tasks when performance fluctuates.

Among these operations, which can be used for troubleshooting and which cannot?

  • A Operations 1, 2, and 3 can be used for troubleshooting, while operations 4, 5, and 6 cannot.
  • B Operations 1 and 2 can be used for troubleshooting, while operations 3, 4, 5, and 6 cannot.
  • C Operations 4, 5, and 6 can be used for troubleshooting, while operations 1, 2, and 3 cannot.
  • D Operations 1, 2, 3, 4, 5, and 6 can all be used for troubleshooting.
Explanation

Workload and I/O-pattern changes, module-level latency, cache-hit-rate changes, very small I/O requests, storage hardware bottlenecks, and background tasks can each cause or correlate with fluctuating all-flash storage performance. Therefore, every listed operation is a valid troubleshooting check.

Community Discussion

No comments yet. Be the first to start the discussion!