Exact Extract — Huawei V3.0 Training Material, PDF Page 15
Global I/O priority ensures stable and low latency.
Kunpeng chip + Service splitting enable faster reconstruction.
Flash-oriented design & disk and controller collaboration ensure stable application performance on the live network.
SAN and NAS share the FlashLink-based global storage pool, which provides global deduplication, compression, similarity-based data reduction, global wear leveling, and global garbage collection.
Huawei OceanStor Dorado uses disk-controller collaboration to distribute storage-processing tasks efficiently between controllers, intelligent disk enclosures, and SSDs.
Global garbage collection (GC) coordinates storage-space reclamation and data organization across the storage system. By optimizing garbage collection at the storage-pool level, the architecture reduces unnecessary internal operations and helps maintain consistent flash performance.
Reconstruction offload uses processing resources within intelligent disk enclosures to accelerate RAID reconstruction. Rather than transmitting every reconstruction calculation to the primary storage controllers, supported calculations are performed closer to the storage media.
This reduces controller CPU consumption and backend communication overhead, allowing reconstruction to proceed with less interference to production workloads.
End-to-end read/write priority ensures that foreground application I/O receives appropriate processing priority. Background activities such as garbage collection and reconstruction must be scheduled without unnecessarily delaying critical application requests.
The FlashLink architecture also supports intelligent caching, large-block sequential writes, and metadata optimization.
Although OceanStor Dorado supports controller-level load balancing, load balancing offload is not identified as one of the specific disk-controller collaboration functions in the referenced material.
Verification correction: The question-bank explanation includes load-balancing offload without a corresponding V3.0 source statement, while the training material explicitly identifies foreground read/write prioritization.
Therefore, the source-supported answer is B, C, D . The precise scope of option D refers to global foreground I/O priority rather than a separately specified hardware offload engine.
Review Note: The reviewed disk-controller collaboration answer includes global garbage collection, reconstruction offload and foreground priority; load-balancing offload is not explicitly supported by that slide.