Shulou(Shulou.com)11/24 Report--

Kevin Kelly, author of out of Control, judged: "No matter what business you are in, your business is now a data business." In a word, data has become the cornerstone of today's business society and economic development. All aspects of production, research and development, management and sales services of enterprises are increasingly inseparable from data, and the acceleration of the integration of data and reality brings unlimited possibilities for the development of the industry.

There is no doubt that a new era of the data era has begun. IDC predicts that the global data circle will grow to 175ZB by 2025, with an average annual growth rate of 30 per cent. With the rise of AIGC, data has become the most important factor of production, and AI is the best tool to release the value of data. The integration of the two brings unlimited possibilities for industrial development, and also brings unprecedented challenges to storage, which is the digital base that carries data.

Under the catalysis of AIGC, the storage industry is ushering in a technological change. Guan Tianshu, General Manager and Chief Product Manager of Storage Product Line, believes that "the future storage must be reconstructed in order to better adapt to the new data storage needs of users in the AI era", and the comprehensive reconstruction of architecture, experience and value is the key to realize the new development of the storage industry. Based on the research and exploration of smart storage technology, Xinhua Group has taken the first industry-leading step in the reconfiguration of storage products.

Guan Tianshu, General Manager and Chief Product Manager of Storage Product Line of Xinhua third Group

Why does storage need to be refactored?

Reviewing the development of storage in the past two decades, from basic storage before 2006 to functional storage in 2005-2015, to intelligent storage in 2016-2025, there are profound changes in the demand for data storage services behind each storage reconstruction. demand change has become the biggest driving force of storage reconfiguration in recent years.

In recent years, with the continuous growth of the amount of data and the enrichment of business applications, the underlying storage system is becoming more and more complicated. Especially with the rise of AIGC and the emergence of a large number of applications based on the LLM model, it not only brings a new data generation model, but also puts forward new challenges in data type, data processing, data management and so on. Guan Tianshu stressed that the application based on the LLM model is different from any previous application, and it will have a subversive impact on the capacity, performance, reliability and management of data storage.

First of all, the large LLM model begins to move from text type to multi-mode, and gradually integrates a large amount of audio and video data, coupled with process data, the demand for storage capacity of large model applications will continue to grow in the future; secondly, the accelerated integration of LLM large model applications and scenarios will not only generate a large number of reasoning requirements, but also generate a large number of reasoning requirements. Ultra-large-scale parallel processing and complex IO features require high performance of data storage. Third, the application of large model, as an application with a high degree of engineering, has different CheckPoint in the training stage because of the important characteristics of the data core, which determines that the underlying storage needs to be stable and reliable, and needs to be better adapted to GPU in order to fully release the potential of the large model.

In the face of the rise of AIGC applications, the current mainstream traditional storage systems generally feel tremendous pressure. In order to fully release the value of data, it is inevitable for storage to accelerate reconfiguration. As the leader of digital solutions, Xinhua Group newly released H3C / HPE Alletra MP global decoupling NVMe intelligent all-flash storage, H3C UniStor X10000 G6 series distributed intelligent fusion storage, taking the lead in giving the answer to storage reconfiguration.

How should storage be refactored?

Martin Fowler, a world master of software development, believes in "refactoring" that any fool can write code that a computer can understand. Only writing code that is easy for human beings to understand is a good programmer.

Similarly, in the field of data storage, storage products are excellent products that meet the needs of the future only if they make people use them more easily, conveniently and efficiently. However, the storage system is a system-level product that integrates software and hardware and many complex technologies, which involves a large number of hardware, protocols and media, which means that every reconfiguration of storage is a huge product technological change, which will have a lasting impact on the future market development and user use. With the rise of AI large model applications, the limitations of traditional storage will be rapidly magnified. Therefore, storage must be comprehensively reconstructed from three aspects: architecture, experience and value, in order to meet the new needs of various data services with a new product form.

The first is the refactoring of storage architecture. Storage architecture is the core "skeleton" of storage system, and whether it is robust or not directly determines the quality of storage system products. Reviewing the history of storage development, from dual-live architecture, directly connected matrix architecture to distributed shared cache architecture and all-shared switching architecture, each reshaping of the architecture has promoted the great progress of the entire data storage industry. At present, the architecture innovation of mainstream traditional storage products has been stagnant for many years, but the rapid development of new protocols, new media, new applications and new scenarios makes traditional storage incompetent. In the face of a series of new challenges such as performance and capacity brought by many new applications such as AIGC, Xinhua Group 3 is committed to promoting the reconfiguration of centralized and distributed storage, making use of architectural innovation to adapt to new applications such as AI, so that the storage system can fully release storage performance and provide better management.

H3C / HPE Alletra MP global decoupling NVMe intelligent all-flash storage is a representative work of architecture innovation. Alletra MP's global decoupling architecture implements the industry's first full-live architecture with completely independent control nodes and data nodes, which not only has large-scale vertical and horizontal expansion modes, but also enables independent expansion of control nodes and data nodes. In addition, the control node is unified with the data node, which eliminates the hardware diversity and greatly improves the iterative update speed of the storage system.

The second is the refactoring of the storage experience. In view of the changes of users' own environment, hybrid / cloudy and edge environment has become the standard choice for enterprise digital transformation, and users pursue simple and consistent data services and functions in a complex environment. In recent years, Xinhua Group has continued to promote the concept of smart storage to help industry users get rid of the disadvantages of strong professional and easy-to-use traditional storage. Xinhua 3 shields the underlying complexity of storage through the concept of software definition; at the hardware level, it adopts standard industrial hardware components to reduce the difficulty of hardware system expansion and management; at the same time, it integrates the native concept of AI into the storage system, devotes itself to the creation of an intelligent storage center, and brings users four major storage experience upgrades: extreme performance, extreme reliability, minimalist management, and green and low-carbon storage.

The third is the reconstruction of storage value. As the carrier of data elements, storage should not become an obstacle to limit the release of the value of data elements because of its own bottleneck. Instead, storage should gradually complete the role transformation and become the best partner to unleash the value of data through efficient, reliable, minimalist, green storage solutions. Fully meet the needs of large-scale data storage and fast access in the AI era, support the differentiated storage needs of various segments, and help users deeply tap the value of data and release the vitality of innovation.

Guan Tianshu believes that the restructuring of storage by Xinhua three groups from three aspects of architecture, experience and value will help users completely shield the complexity of infrastructure and focus on upper application innovation.

Accelerate the reconstruction and release the value of data

According to the China Digital economy Development report (2023), China's digital economy reached 50 trillion yuan in 2022, accounting for 41.5% of GDP, and industrial digitization accounted for 81% of the digital economy. The digital transformation of various industries increasingly needs advanced infrastructure to support the full release of data value. Just as new energy + intelligence has been reconstructing the automobile industry for a hundred years in recent years, as the digital base of the digital transformation of enterprises, the reconstruction of storage is an embodiment of the evolution of the storage industry itself, and this reconstruction will be a subversive change from the inside to the outside. its core goal is to make it no longer difficult to release data value.

As the leader of digital solutions, Xinhua three groups comprehensively promote the landing of storage reconfiguration from three aspects of products, scenarios and concepts, build advanced infrastructure, and help users in various industries to complete the digital transformation.

From the product level, Xinhua Group took the lead in launching H3C / HPE Alletra MP and H3C UniStor X10000 G6 two AI native architecture storage. The core goal of Alletra MP is to help enterprise customers restructure their key capabilities to meet a series of data storage challenges from edge to cloud scenarios, such as complex management, isolated infrastructure and inefficient expansion; X10000 G6 is positioned as a solid digital base in the AI era, focusing on the support of various AI applications. The two products are positioned differently and well cover the storage needs of users' digital transformation.

From the scene level, Xinhua three groups around the differentiated needs of different scenarios, according to product characteristics to promote the landing of storage reconfiguration. X18000 G6 can fully meet the full scene storage needs such as AI, virtualization and database, and is very suitable for AI scenarios such as scientific research, AI painting, self-driving and so on. While Alletra MP, with high reliability, excellent performance, unified platform and consistent cloud experience, coupled with AI blessing, is very suitable for various scenarios with structured data.

From the conceptual level, Xinhua Group has been promoting the popularization of the concept of smart storage. Improve and upgrade storage products through AI technology, realize the perception of changes at the application level, grasp and understand business rules, and make correct analysis and judgment, so as to make data storage operation and management easy. Xinhua San hopes to bring the concept of smart storage to more users and scenarios through two new storage products, Alletra MP and X10000 G6.

Facing the future, Xinhua Group has a complete reconfiguration storage system plan to promote the continuous reconfiguration and evolution of storage products. Guided by the "Yunzhi Native" strategy, Xinhua 3 will focus on Yunzhi native technology, deepen customer application scenarios, continuously strengthen technological innovation, and constantly launch more smart storage products to help users of all industries eliminate data storage challenges, accelerate the release of the value of data elements, and jointly move towards a new era of AI.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.