Exact Extract — Huawei V3.0 Training Material, PDF Page 394
Under the same test conditions, using the same controller failure simulation method, observation dimension, and I/O model, you are advised to observe test results on the host side.
Huawei OceanStor Dorado uses SmartMatrix's fully interconnected architecture to maintain storage-service availability during supported controller-failure scenarios.
When comparing storage reliability across vendors, a POC must use consistent testing procedures to ensure that the results represent differences in the underlying architectures rather than differences in test conditions.
Option A: The same observation dimension must be applied to all systems. For example, host-side IOPS, latency, and response-time measurements should be collected consistently throughout the test.
Option B: Host-side monitoring is recommended because individual storage vendors may use different internal sampling intervals, averaging methods, and performance-reporting mechanisms. Host-side measurements provide a consistent view of the application-visible effects of controller failures.
Option C: The same I/O workload must be used. A mixed read/write workload is particularly useful because controller failure can affect reads and writes differently. Identical block sizes, queue depths, read/write ratios, and concurrency levels improve comparison validity.
Option D: Controller failures should be simulated under equivalent conditions, with the same workload and observation methods. Physical controller removal should be performed only according to an approved test procedure and the equipment's supported maintenance requirements.
These measures help evaluate service continuity, transient latency, performance degradation, and recovery behavior.
Therefore, all four options are correct .