Exact Extract — Huawei V3.0 Training Material, PDF Page 220
Efficient Write Cache Design and Three Key Technologies for Improving Small I/O Write Performance
When DRAM is used as the cache, the I/O operation latency is 1% of that of SCM and 1‰ of that of SSDs.
Unstructured services use the innovative BBU + DRAM architecture.
The RDMA technology is supported for cache mirroring between nodes.
Huawei OceanStor Pacific uses an optimized write-cache architecture to improve the performance of small random I/O operations.
The V3.0 training material identifies three principal technologies.
First, DRAM-based caching reduces write-response latency. DRAM offers substantially lower access latency than persistent storage media, allowing small I/O requests to complete quickly after the required cache-protection conditions are satisfied.
Second, the BBU + DRAM architecture combines high-performance volatile memory with battery-backed power protection.
This provides fast write caching while helping protect cached data during supported power-failure scenarios.
Third, RDMA-based cache mirroring enables data to be mirrored between storage nodes with reduced CPU intervention.
Huawei identifies a latency reduction exceeding 50% for the corresponding cache-synchronization optimization in the documented comparison.
Option B, larger cache space, is not one of the three explicitly identified technologies.
However, this distinction creates an issue with the question.
Using a larger cache is not inherently an incorrect design practice, and the training slide does not establish that OceanStor Pacific cannot use additional cache capacity.
The source-supported distinction is that the three highlighted mechanisms are DRAM caching, BBU-protected DRAM, and RDMA cache mirroring.
Verification correction: Option D must not be marked false. The Huawei training material explicitly identifies DRAM as the cache medium.
The best intended answer is B , but the question should be rewritten as:
Which of the following is NOT one of the three key small-I/O write-performance technologies explicitly identified in the OceanStor Pacific V3.0 training material?
That wording makes B defensible without claiming that larger cache capacity is universally incorrect.
Review Note: B is the best exam answer as "larger cache" is not one of three technologies listed; larger cache is not inherently incorrect.