From lightweight network to data storage, Huacun Zhigu collaborates with Huawei to release two major solutions

At the Huawei Partner Conference 2025 held in Shenzhen on March 20-21, 2025, Huacun Zhigu, as a dual strategic partner of Huawei in data storage and data communication, jointly released two innovative solutions with Huawei, comprehensively demonstrating its technological strength and industry practice achievements in the fields of data storage and data communication.

On the afternoon of March 20th, Chen Hao, Senior Product Architect at Huacun Zhigu, gave a keynote speech on "Data Intelligence Integration: AI enabled Data Storage Center Architecture" in Public Lecture Area 4.

Challenges and Opportunities of Data in the AI Era

The current AI big model technology is flourishing, but the implementation of general big models in vertical industry fields faces core challenges such as poor data quality and insufficient diversity. The data in vertical industry fields is scattered and isolated, with unstructured data accounting for 90% of the total. The knowledge conversion rate of unstructured data is low, making it difficult to form domain specific knowledge to support the training of domain specific large models. Building a new data architecture that integrates multimodal data access, global view, and knowledge extraction capabilities has become the key to promoting high-quality development of AI.

Innovative architecture and technological breakthroughs

Huacun Zhigu, in collaboration with Huawei and Peking University Clamshell Intelligence, has launched the "Data Storage Center Architecture" to achieve full stack capabilities for data aggregation, supply, management, and usage, promoting data aggregation and utilization. This architecture is based on a global data lake, which enables the aggregation and high-speed access of cross domain heterogeneous multimodal data; Combining the knowledge platform to achieve AI driven data governance and knowledge extraction, empowering the intelligence of industry scenarios. The core technology includes: independently developed high-performance parallel file system that balances performance and utilization, 91.6% high storage utilization, industry-leading peak performance (single device 175GB/s bandwidth, 7.5 million IOPS); Full functional data access interface support, data sharing and interoperability across protocols, and zero latency flow of data; Global metadata service, building a unified global data management view to achieve cross layer, cross domain, and cross business data circulation across "hot, hot, cold" layers; The knowledge platform enables data decoding through AI, realizing professional knowledge extraction and systematic governance, and building a domain specific knowledge base for enterprises; By customizing fine-tuning and mixing expert models, we aim to create an enterprise exclusive cloud brain that enables intelligent business scenarios.

Industry practice and value realization

With the rapid development of the AI industry in the past two years, this innovative architecture has shown a multi-point flowering situation, including the landing of general scale scenarios such as domestic Wanka clusters and regional data storage centers, as well as the landing of vertical professional fields such as finance. Huacun collaborates with partners to train multiple large-scale models in a 10000 card cluster project, fully supporting the efficient operation of computing power clusters with TB level aggregated bandwidth; The regional data storage center is currently under construction, with three computing power centers in the entire region serving heterogeneous computing power clusters with homogeneous/heterogeneous data sources, achieving mutual promotion of data element circulation and computing power resource consumption, and using data aggregation to promote computing; In a case study of a city commercial bank, we integrated over 80 sets of business systems with billions of unstructured data, built a domain knowledge base, and enabled rich intelligent agent applications to form a "data knowledge intelligent agent" closed loop, helping enterprises achieve a comprehensive transformation from data element operation to intelligent transformation.

Chen Hao, Senior Product Architect at Huacun Zhigu

On the afternoon of March 21, at the Data Communication Sub-Forum themed "Galaxy AI Data Center Network, Empowering New Momentum in the AI Era," Wu Yingqiao, Network Technology Director of Huacun Zhigu, released "Huacun Zhigu and Huawei Jointly Create a Lightweight Solution for the Commercial Data Center Market."

Solution Background and Value

With the acceleration of digital transformation, SMEs face challenges such as high data center construction costs and complex operation and maintenance. Huacun Zhigu, in collaboration with Huawei, has launched a lightweight data center solution. Through a simplified architecture, intelligent operation and maintenance, and elastic scalability, this solution helps customers quickly build efficient and stable data centers, significantly reducing technical barriers and total cost of ownership (TCO).

Core Technological Advantages and Innovations

The solution relies on a triple-guarantee system built on equipment reliability, network reliability, and operation and maintenance reliability to comprehensively support SMEs in achieving digital transformation. For AI scenarios, the innovative application of NSLB technology improves AI training efficiency by 40%. Combined with Starlink optical modules featuring channel resilience technology, it provides high-performance network support for computing platforms such as Ascend, meeting the diverse needs of SMEs transitioning to AI intelligent computing centers.

Wu Yingqiao, Network Technology Director of Huacun Zhigu, further solidified the collaborative innovation model between Huacun Zhigu and Huawei in the fields of data storage and communication with his presentation at the Huawei Partner Conference 2025. In the future, both parties will continue to deepen cooperation in two main areas: continuously optimizing the "integrated storage and intelligence" architecture, focusing on streamlining the technical chain from multi-source data aggregation to AI knowledge extraction; promoting deep adaptation of lightweight solutions to the Ascend computing platform; exploring a standardized service framework for the linkage of AI networks and storage; and providing industry customers with more cost-effective end-to-end solutions.