The correct answers are A, C, and D . In IVS3800 architecture, the distinction between management nodes and service nodes is based on whether the node carries the platform’s management container set, especially the CSPOM role. A management node is the server that contains the management component, while a service node is the server that does not contain that management component and instead focuses on service workloads. This is why C and D are correct from the node-role definition perspective.

The cluster nature of the management role is also consistent with the platform design. The training material repeatedly refers to BMU and management addressing in two-node cluster mode , for example: “If the BMU is deployed in two-node cluster mode, enter the external service floating IP address of the BMU” and similarly references external management floating IP handling for clustered upper-level nodes . This reflects that the management layer is designed around clustered high-availability deployment, supporting A . By contrast, B is not accepted because an IVS3800 system made up of only management nodes would lack the service-bearing nodes needed to deliver actual intelligent vision workloads. Therefore, A, C, and D are the correct statements.

The physical design of a data center for intelligent vision systems involves strategic placement of network switches to optimize cabling and performance. Two common architectures are Top of Rack (TOR) and End of Row (EOR). In a TOR deployment, Top of Rack (TOR) switches are deployed on the top of each cabinet and used for communication between servers in each cabinet and the external network . This approach minimizes the length of copper cabling within the rack and simplifies the expansion of individual cabinets.

Conversely, the EOR architecture centralizes the switching infrastructure. In this setup, End of Row (EOR) switches are deployed in the last cabinet at each row, and servers in all cabinets at each row are connected to switches through horizontal cabling . While EOR can simplify management by centralizing hardware, it requires significant horizontal cabling across the row. Huawei’s IVS planning recommendations specify that the management port and service ports of hardware devices (like the Taishan 200) connect to these switches. Specifically, the "Management TOR" handles the management plane (often using CE5800 series), while the "Service TOR" handles the high-bandwidth 10 GE service plane (often using CE6800 series), with both eventually aggregating through the EOR switch to connect to the wider customer network.

1 + N → Sends video streams from generic cameras to intelligent cameras and performs real-time intelligent analysis on the video to embed intelligence into generic cameras.

T-shot engine → Uses framing technology to detect targets and license plates on images at the same time, improving the snapshot accuracy at night.

SuperColor → Helps capture full-color images in low light conditions.

Scene adaptation → Automatically identifies various scenes such as backlight, low light, and rainy/overcast scenes and optimizes images accordingly.

The matching is determined directly from the intelligent camera feature definitions. The 1 + N function is the primary-secondary camera observation capability, where “a single intelligent camera can connect to N generic cameras on a live network, request streams from these cameras, and perform intelligent analysis on the streams”

so it matches the description about embedding intelligence into generic cameras.

The T-shot engine corresponds to the snapshot-related description because the material states that “M-Q and X-H series omni-data structuring cameras support T-Shot snapshot engine” and explains that “T-Shot, Huawei's patented exposure technology, enables optimal imaging of targets and vehicles at the same time” . That fits target and license plate capture in the same image.

Scene adaptation matches the description about identifying backlight, low light, and rainy or overcast scenes because the material states that AI chips “identify the weather and illumination conditions at different time points in the observation scenario, and then automatically adjust the image settings”

The remaining feature, SuperColor , therefore matches full-color imaging in low-light conditions.

The correct answer is A, B, C, and D because the Kunpeng computing industry is an end-to-end ecosystem rather than a single hardware product category. In the material, Kunpeng is presented not only as a processor platform, but as part of a broader ecosystem. The chip layer is evident from the discussion of processor architecture and capabilities, while the software layer is reflected where the document notes that “Kunpeng products use the following mainstream OSs: CentOS, Ubuntu, NeoKylin, Debian, Huawei EulerOS, Kylin, SUSE, and Deepin” . That directly confirms software and operating-system support as part of the industry foundation.

The hardware server layer is also intrinsic because Kunpeng processors are deployed through server products and computing infrastructure, not in isolation. On top of hardware and software, applications complete the industry chain by delivering business value in cloud, AI, and intelligent vision scenarios. In industry terminology, a computing ecosystem normally spans chips, hardware platforms, basic software, and applications , and Kunpeng follows that same structure. Therefore, all four options describe valid components of the Kunpeng computing industry, making A, B, C, and D the correct answer set.