Solve technical problems in key areas! Science and technology enterprises have entered the wave of domestic self-research

Today, the momentum of domestic substitution has been "overflowing". In recent years, high-tech enterprises have been included in the "entity list" from time to time, and the situation of "stuck neck" in key areas still exists. Under the background of the era of big power game, the importance of independence, safety and controllability of technologies and products in key fields has been continuously highlighted in international competition.

Fortunately, a number of domestic science and technology enterprises have entered the wave of research and development of domestic alternatives and solving technical problems in key fields. These enterprises concentrate on industrial manufacturing, information industry and other industries, breaking foreign technology monopolies in fields related to national security, manufacturing, national economy and people's livelihood, shortening the technological gap, strengthening the moat of domestic technology, promoting high-level self-reliance and self-improvement in mainland science and technology, escorting the mainland's social and economic development, and becoming a unique "national tide" in scientific and technological development and national competition.

360

360GLM large language model: provides a solution idea for the research and development of "domestic ChatGPT"

With the advent of ChatGPT, large language models have become a key development direction in the field of artificial intelligence, and the R&D competition is unprecedentedly hot. The industry generally believes that large language models will bring an industrial revolution driven by strong artificial intelligence or general artificial intelligence, but the objective situation is that the mainland is still catching up in the field of large language models. Zhou Hongyi, founder of 360 Group, once reflected in an interview with the media that in the past, the research and development of artificial intelligence in China was too pragmatic. However, Zhou Hongyi also believes that in two or three years, China has the opportunity to catch up in terms of big language models.

360 provides a possibility for the field of large language models to catch up, and also provides a solution idea for the research and development of "domestic ChatGPT".

On May 16, 360 and Zhipu AI announced a strategic cooperation, and the 100-billion-level large model "360GLM" jointly developed by the two parties has reached the level of a new generation of cognitive intelligence general model. The joint R&D between the two parties is an effective combination of the basic model and the application scenario. Zhou Hongyi has said many times that Microsoft, as an industrial company, has completed OpenAI's engineering, scenario, productization and commercialization capabilities, and it is the division of labor model between Microsoft and OpenAI that has contributed to the emergence of this artificial intelligence inflection point. China should establish a collaborative innovation model of production and research of large science and technology enterprises + key scientific research institutions, and create China's "Microsoft + OpenAI" combination to lead large-model technology research. He said that the cooperation with Zhipu AI draws on this industry-research synergy.

According to reports, Zhipu AI was evaluated as the "most OpenAI temperament and level" artificial intelligence company in China. According to the evaluation of the Large Model Center of Stanford University, the accuracy of the bilingual 100-billion-level ultra-large-scale pre-training model GLM-130B developed by Zhipu AI is close to or equal to the large models of OpenAI, Google Brain, Microsoft and NVIDIA. 360 has a domestic advanced multimodal R&D team, which has long-term accumulation in artificial intelligence technology, superimposed search, browser and other scenario advantages, which can provide useful supplements in training data, reinforcement learning, engineering tuning, user scenarios and commercialization.

It is understood that through this cooperation, 360 has formed a large model layout driven by independent research and development + cooperative research and development "dual engine". With the close cooperation of the two parties, domestic large-language model technology will obtain more extensive and in-depth landing scenarios and empower more industries.

Tencent

Financial Cloud: Build a full-stack leading domestic digital base based on products

Not only databases, cloud platforms and other basic software are also growing rapidly in technological innovation and industrial applications, and have become a unique existence in some fields.

For example, in the financial industry, basic software is monopolized by foreign giants and financial institutions face cost problems. As a "reservoir", the financial market is crucial to the flow of funds and the national economy and people's livelihood, so it is particularly important to be self-reliant and self-reliant in technology and software related to the financial industry. In addition, the central bank and the Financial Supervisory Authority have also issued guidelines for the development of financial technology, emphasizing autonomy and control in key product technologies.

Fortunately, the basic software of the financial industry has completed the localization of replacement, and now it has entered the deep water area of localization replacement of core systems.

On May 11, at the Tencent Financial Cloud Localization Strategy Summit, Hu Liming, Vice President of Tencent Cloud, officially released the financial localization strategy: Tencent Cloud will continue to increase investment in self-developed products, work with financial industry partners to improve the localization ecology, form independent and controllable financial localization solutions, and build a full-stack leading domestic digital base based on products. At the same time, Tencent Cloud will work with partners to build a core solution for financial distribution.

On the technical foundation, Tencent Cloud provides products including: Tencent Proprietary Cloud TCE, Financial Native Cloud TCS, Tencent Cloud Database TDSQL, TDMQ Distributed Message Queuing, CRedis Distributed Cache, Coding R&D Performance Platform and other products, which fully cover the technical foundation required for the distributed core system of banks, and have been successfully delivered in the construction of the core systems of dozens of banks such as state-owned banks, joint-stock banks, city commercial banks, and provincial rural credit unions, and are mature and stable.

Among them, Tencent's proprietary cloud TCE has built a cloud multi-core domestic technology base. Based on Tencent's proprietary cloud TCE, financial institutions can quickly build a financial-grade ecological cloud platform that meets national standards.

Hu Liming introduced that Tencent Financial Cloud's full-stack domestic product supply capacity has achieved national leadership, and will continue to increase technology investment in the future to create a new generation of domestic quality products.

It is reported that the supply capacity of Tencent Financial Cloud full-stack localized products has achieved national leadership, and the underlying basic products have completed the localization adaptation with many chip manufacturers. Up to now, a total of 15 products have passed 21 quality tests of the Ministry of Industry and Information Technology, and more than 700 independent innovation mutual certifications have been completed with partners.

Tencent Cloud said that Tencent Financial Cloud will continue to cultivate the localization track, improve the core competitiveness of products, and work with ecological partners to help the digital transformation of the financial industry out of the deep water area.

Ant Group

OceanBase: Let domestic databases go to the field

The database is related to the national economy and people's livelihood. Every moment of the operation of society will generate massive data, and the storage, management and retrieval of massive data are inseparable from databases. However, in fact, the technical strength of the domestic database field has been relatively weak for a long time, which makes the database one of the 35 key technologies of the country's "stuck neck".

In this context, the field of domestic databases has been striving to break the monopoly of foreign technology and test the waters to break through new technologies. It can be seen that at present, the domestic database has matured and begun to try to go general, test the water and open to ecology.

On March 25, the first OceanBase Developer Conference was held in Beijing. OceanBase 4.1 was released, announcing two friendly tools, upgraded documentation usability, and unified enterprise and community edition code branches, fully demonstrating OceanBase's efforts to build the ultimate developer-friendly database.

OceanBase Database Appliance. It provides in-depth performance tuning of software and hardware for database services, providing high reliability, high security, high cost performance, intelligent management and control, and one-stop rapid delivery and deployment capabilities.

Meanwhile, OceanBase unveiled four "developer-friendly" practices for the first time. First, OceanBase version 4.1 was released - transaction processing performance increased by 40%, data analysis performance increased by 15%, bypass import improved data loading performance by an average of 6 times, new tenant-level physical backup databases and more fine-grained resource isolation capabilities, fully compatible with MySQL 8.0, more versatile and easy to use. Second, the introduction of wizard-style installation and deployment lowers the first threshold for using distributed databases. The third is to introduce scenario-based documents to solve the problem that documents are "difficult to find and difficult to use". For a long time, domestic database technical documentation has dissuaded a group of database enthusiasts. Ocean-Base combined the problems and suggestions from the outside world, designed from the user's perspective and scene, and reconstructed more than 7,000 pages of documents. Fourth, it announced that the entire R&D process will be migrated to the outside, unifying the enterprise version and the community version code branch, and announcing that the three major developer tools will be open source, taking an important step towards ecological openness.

Yang Chuanhui, CTO of OceanBase, said at the conference: "OceanBase will continue to lower the threshold for developers, comprehensively improve the ease of use of OceanBase, create a truly developer-friendly database, build an open technology ecology, and let domestic databases go to the field." ”

It is understood that up to now, OceanBase has achieved industry-wide application scenario coverage, helping more than 400 customers in finance, government affairs, operators, retail, Internet and other industries to achieve key system upgrades. Yi Hongwei, senior R&D director of Ocean-Base, said: "Ocean-Base has moved from finance to general use, breaking the stereotype that distributed databases can only support large enterprises and large customers in the past, and can better serve small and medium-sized enterprises. "Text/Li Mei

OceanBase won the leading scientific and technological achievements of the 2022 World Internet Conference, and Yang Zhenkun received the award at the scene.

Yang Zhenkun, founder and chief scientist of OceanBase:

"Changing lanes and overtaking" in the international competition of databases

Living in such an era of mobile Internet, everyone is producing a large amount of "data" every day. Our travel records, consumption records, web pages viewed, and messages sent are data, as are the music we hear and the videos we see. Usually, this huge and complex data will be stored in a warehouse called "database" according to certain rules.

For more than half a century, the global database industry has been in a relatively stable state in terms of technology and applications, and its market share is mainly in the hands of several European and American companies such as Oracle (Oracle), as well as the Chinese market. Relevant data show that the global database market size in 2020 was US$67.1 billion, of which the Chinese database market size was US$3.5 billion (accounting for 5.2% of the world), which was almost monopolized by the databases of foreign brands.

To illustrate this situation in a very graphic metaphor, it is that in the past few decades, we have stored a lot of important information and data in data "warehouses", but in fact the keys to these warehouses are not in our own hands. It wasn't until 2010 that Yang Zhenkun's new attempt began, and the Chinese database market began to have some new changes.

Yang Zhenkun is the vice president of Ant Group, and the founder and chief scientist of OceanBase. In 2010, 45-year-old Yang Zhenkun began to lead the team to completely independently develop the distributed relational database OceanBase. After more than ten years of independent innovation, OceanBase has been authorized 202 invention patents, applied for more than 350 invention patents, 19 software copyrights, published more than 40 papers, led and participated in more than 25 national/industry/group standards, and published 1 monograph Chinese. The 3 million lines of core code have been based on the Mulan license agreement and opened to the outside world using the OpenCore model, and are committed to building the technical route and ecosystem of database products with their own property rights in mainland China.

In the various "actual battles" experienced, OceanBase has steadily supported "Double 11" for 10 consecutive years, innovatively launched a new standard of "three regions and five centers" city-level disaster recovery, and twice broke the world record of "Database World Cup" TPC-C in 2019 and 2020, becoming the first distributed relational database on the list. At present, OceanBase has become the first choice for upgrading financial-grade database core systems.

According to public data, at present, OceanBase has been applied to more than 1/4 of the domestic leading financial institutions, and has begun to move from finance to the national economy and people's livelihood and overseas, and has now served Industrial and Commercial Bank, China Construction Bank, Bank of Communications, Shanghai Pudong Development Bank, Chinese Life, Chinese Insurance, China Property & Casualty Insurance, China Merchants Securities, Shandong Mobile, Jiangsu Mobile, Sinopec, Haidilao, Lili, GCash, etc., to help government agencies, finance, operators, retail, More than 400 customers in various industries such as the Internet have achieved key system upgrades.

In the eyes of industry insiders, the successful R&D and large-scale application of OceanBase verifies the feasibility of domestic database technology in "changing lanes and overtaking", provides a new track and space for international competition in the database field, not only lays a solid foundation for the transformation of distributed architecture in various industries, but also contributes to the security and information security of information infrastructure in key industries of the country, and also provides strong scientific and technological support for the national "dual carbon" strategy.

When the "Zhengguochao" planning was in full swing, we interviewed Yang Zhenkun, who claimed to be the "oldest programmer of Ant Group". His openness and unique insights on many issues have given us a whole new understanding of the database industry.

Beijing Daily: Why does OceanBase focus on domestic self-developed databases? What are the needs of domestic databases for the development of domestic markets and industries?

Yang Zhenkun: The relational database is the key information infrastructure of the entire society, supporting the entire national economic and social development. The rapid development of the digital economy has made the relational database from closed environments such as shopping malls, hotels, factories to open environments such as the Internet, traditional centralized databases such as Oracle can not solve the demand for data processing in scenarios such as "Double Eleven", and there is no mature solution in the world, and the distributed relational database OceanBase came into being.

In fact, the development of the domestic market and industry has clear requirements for the cost requirements, scalability and real-time data analysis capabilities of domestic databases. It is generally understood that the cost should be low, and the performance cannot be low. Now, OceanBase has also been well polished in Taobao, Alipay and other business scenarios, and has been widely used in domestic banking, insurance, securities, energy, communications, government affairs and other fields. In particular, in the core system upgrade of financial institutions, OceanBase's performance advantages are particularly obvious.

Beijing Daily: Different from foreign databases, what important technological breakthroughs has OceanBase made in the process of building a domestic self-developed database? What does such a domestically produced self-developed technology mean for Beijing to build a highland of information and innovation industry? What does it mean for the strategic development of the country?

Yang Zhenkun: From the first day, OceanBase has followed a completely independent research and development model, creating the route and technology of native distributed database, as well as the current single-machine distributed technology. At the beginning of research and development, these are all things that others do not have, and they are completely different from others. From a technical point of view, OceanBase is the pioneer and leader of distributed databases, up to today, most of the domestic is still based on middleware distributed or cloud-based shared storage model, they are actually "centralized +" concept, OceanBase has been native distributed, as well as stand-alone distributed. In this dimension, we have broken the technical boundaries of databases.

In fact, China's basic software has entered the stage of capable root innovation, and global relational database products and technologies are moving from centralized to native distributed, and domestic databases are no exception. Domestic databases need to actively participate in international competition and strive to become the global mainstream database.

As the three core "stuck neck" projects, if the database can come out on the highland of database self-research, it will surely help Beijing's related industries, and at the same time, Beijing's vigorous development of relevant policies and measures for basic scientific research will also be conducive to breakthroughs in basic scientific research like databases. This is a linkage model, and it is also an exploration of a new model for the country. We need to have confidence and strategic determination, as long as we adhere to long-term investment, the future domestic database will definitely usher in a big outbreak.

Beijing Daily: In March this year, OceanBase announced three major open source developer tools, taking an important step towards ecological openness. What does open source and ecological openness mean for OceanBase? What is the significance for the development of domestic databases?

Yang Zhenkun: The development of databases, ecology is very core and critical. Among them, talent training is the cornerstone of the development of the database industry. Since its establishment, OceanBase has attached great importance to the cultivation of database talents, and has jointly established joint laboratories and subject research centers with many domestic universities such as East China Normal University and Wuhan University, jointly output papers, patents and scientific research results, and actively participated in the cooperation of domestic database curriculum reform, using the power of enterprises to solve the problem of disconnection between theoretical education and practical work in schools.

In this process, we have developed a MiniOB teaching practice platform and supporting teaching materials and tutorials that specifically exercise students' core development capabilities, and also held two consecutive OceanBase Database Competitions to help the construction of database talent ecology through practical actions.

Open source allows more people to use them, let more people understand and use these tools, so as to innovate on this basis. The biggest advantage of open source is that it can lower the threshold of learning, provide a shared learning atmosphere and test scenarios for engineering practices, and build a healthy and prosperous ecosystem. In June 2021, we officially open sourced more than 3 million lines of code, using China's Mulan open source license. Since open source, OceanBase Community Edition has received many positive reviews from developers and industry users, and has been tried and used in the production environment of Internet, banking, insurance and other industry users, and more than 300 customers have carried out in-depth practice, including Ctrip, Zhihu, Kuaishou, Yiou Education, Homework, Yipay, Pudong, 360, 58.com and other customers.

OceanBase has connected more than 30,000 community developers, and since October 2022, OceanBase has ranked first on the Chinese database popularity list Motianlun for six consecutive months, and ranked 126th in the global database popularity list DB-engines, ranking at the forefront of domestic databases. Only when more and more enterprises participate in the ecology of self-developed databases, the database industry will flourish more.

OceanBase is building a partner-centric business model and looks forward to in-depth cooperation with more industry ecosystems to jointly promote the development of the database industry.

Beijing Daily: What are your expectations for OceanBase's development?

Yang Zhenkun: OceanBase wants to make a technology that "stands above the sky". "Topping the sky" means constantly pursuing new breakthroughs in technology, and "standing on the ground" is to make technology into a general product so that the whole society can use it universally. Ocean-Base was born at the right time and place, which is a once-in-a-lifetime opportunity. The database industry is a research and development, long-term matter, OceanBase will continue to work hard for Chinese databases to gain a foothold in the world. Let the self-developed distributed database break through the technical boundary, go to the top of the world, and land in thousands of industries, we must do, and we will definitely be able to do it.

Text/Liu Yang

(Original title: Domestic self-research: moving forward in the wave)

Source: Beijing Daily

Process Editor: U016

Copyright statement: The copyright of the text belongs to Beijing News Group, and may not be reproduced or adapted without permission.