On April 11th, the 2024 Huawei Financial Innovation Data Infrastructure Summit with the theme of & ldquo; Data innovation future, smart finance & rdquo; Was held in Chongqing. At this summit, Zhou Yuefeng, president of Huawei's data storage product line, officially released four major solutions for the financial industry: AI, edge AI, AI data protection, and AI data Lake, help the financial industry embrace AI and build a reliable data infrastructure in the AI era.

Dr. Zhou Yuefeng, vice president of Huawei and president of data storage product line

the rapid development of AI models has brought about changes in service scenarios and business models in the financial industry, and the value of financial data assets will be fully released. In this process, the construction of data infrastructure in the financial industry faces three major challenges:

first of all, the scale of AI large model clusters has entered the era of 10,000 kcal and 100,000 kcal. Large clusters bring more frequent cluster failures. Repeated data preparation and resumable training lead to idle computing resources, the cluster utilization is less than 50%. In addition, the AI model contains a large amount of high-value data, which is more likely to be the target of ransomware attacks or sample poisoning, and has great data security risks. Finally, lack of data and AI, how to manage and make good use of data assets has become a prerequisite for AI development.

To meet the above challenges, Huawei has released a series of innovative AI solutions to help financial customers build a secure, reliable, and open AI data infrastructure for multiple scenarios such as Central AI and edge AI, enable business innovation in the financial industry.

| facing the AI scenario of the financial industry center, Huawei has launched the AI solution of the Center: build an AI Cluster System (ACS) by integrating the one-stop data processing tool chain framework edatespec, model-enabled tool ModelMate, application development interface framework AppEngine, NPU virtualization, containers, shengteng computing power enables software CANN, MindX and high-performance OceanStor A series of professional AI storage to give full play to the collaborative advantages of storage computing network, realize unified management and maintenance, and effectively improve cluster availability by 30%, create an optimal system-level solution for AI centers. The central AI cluster system is extremely open. Open support for third-party data processing, open support for hundreds of models, and empower AI application development partners.

| facing the edge AI scenario in the financial industry, Huawei has launched an edge AI solution.: the built-in data processing tool chain framework of FusionCube A3000 all-in-one training/push-and-super integration machine, edatespec, enables fast collection of multivariate data and accelerates knowledge generation. Embedded RAG can update inference business knowledge in real time and eliminate the illusion of big models, meet the requirement of precise inference of tens of billions of parameters; Carry DME management software to manage basic hardware and software platform resources such as storage and computing network as a whole; Preset computing power enabling software such as AppEngine and mindset, accelerate the launch cycle of AI application development. Huawei FusionCube A3000 actively cooperates with various model manufacturers to accelerate the commercial realization of AI technology.

| for AI data protection scenarios in the financial industry, Huawei has launched a secure and high-performance AI data protection solution: by deploying the DME data management engine, you can accurately identify various risks in AI clusters and implement precise security policy management for AI clusters. By deploying OceanCyber data security cards in storage clusters, perform entropy analysis on the written data to realize the detection and analysis of the whole process of ransomware attack in advance, in the process and after the event. The detection accuracy of ransomware attack is as high as 99.9%. By deploying OceanProtect all-in-one backup machine, efficiently back up key data to achieve second-level data recovery, tamper-proof and ransomware-proof, and ensure that AI training data is highly reliable and business is not interrupted.

| facing the challenges of Enterprise global data management, Huawei launches AI data lake solution: the storage cluster capability in AI cluster system (ACS) is enhanced. Based on the intelligent grading capability of OceanStor A series of storage clusters and OceanStor Pacific series of mass storage clusters, DME provides the GFS global file system capabilities to make enterprise data assets visible, manageable, and available.

Chen Lin, general manager of Huawei's financial system department in China, said that AI is gradually being applied in the financial industry, and a solid and flexible data base is increasingly becoming the cornerstone of AI innovation. Huawei is committed to building AI big model application, data security and data management capabilities in the financial industry through the accumulation of data storage industry, and works closely with customers and partners to jointly build a safer, A stable and efficient financial resources ecosystem helps the financial industry build innovative data infrastructure.

AI provides power for innovation of financial services. Huawei data storage builds an AI cluster system oriented to central AI, and a FusionCube A3000 training/push-and-super integrated machine oriented to edge AI, A comprehensive AI data protection solution and an AI data lake solution that enables global data assets to be visible, manageable, and available help Financial customers build AI-Ready data infrastructure.