A Gift to the Society: Alibaba’s Road of Technological Innovation
By Huang Xianliang, Marketing and Public Affairs Department, Alibaba Cloud
Editor: Zhang Nan, Alibaba Cloud Research Center
Scientific and technological innovation capability is the core of national competitiveness and the source of enterprise competitiveness. As a world-class scientific and technological enterprise driven by scientific and technological innovation, Alibaba actively participates in the construction of national scientific and technological innovation system, by highly coupling enterprise strategies with national strategies. Alibaba has deeply deployed in the bottle-neck and disruptive technology fields such as operating systems, cloud computing and databases, and has achieved a series of important results.
Overview of Alibaba’s Technological Innovation
Alibaba Group has maintained a high level of investment in research and development. Its investment in technology and research and development in recent years exceeds RMB 100 billion per year. As of March 31, 2020, Alibaba Group has obtained about 8,000 Chinese patent licenses, 14,000 open patent applications, 4,000 applications for foreign patent authorization，12,000 foreign open patent applications.
Alibaba’s Technological Innovations
1) Cloud Computing
Since writing the first line of codes in 2009, Alibaba has embarked on the road of independent research and development of China’s cloud computing and developed the Apsara system.
As China’s unique, self-developed computing engine, the Apsara system is a super-complex project. The total codes developed by Apsara stack exceed 2 billion lines, which obtained the first in 15 years, which is the Grand Prize for Scientific and Technological Progress awarded by Chinese Institute of Electronics.
The Apsara system creatively puts forward the architecture principle of “data-centric cloud computing”, which adopts the technical route of global aggregation of resources and then logical segmentation. It achieves global management, unified scheduling, fast positioning, and timely access in a globally distributed data center environment. It connects millions of servers to a supercomputer for the first time.
Alibaba Cloud supported by the Apsara system ranks first in China’s public cloud market share. It supports Alibaba Cloud to build China’s own universal computing technology platform, such as Systemanalyse Programmentwicklung (SAP), Siemens Mindsphere, Salesforce, and Vmware. Some of the world’s most important software run on it. The incubation has cultivated innovative applications such as City Brain, Industrial Brain, and Agricultural Brain, which builds a batch of industry innovation solutions.
At the Apsara Conference on September 17, 2020, Alibaba showed the first cloud computer, “Wuying”. This is a “super computer” on the cloud, no local host. All hardware devices are located in the cloud data center. You only need a C-Key (the size of a card holder) to access applications and files, carry out design modeling, animation rendering and other operations like the traditional PC. In addition, unlimited scaling through the cloud can be implemented, and single application resources can be elastically scaled to 104-core CPU, 1.5 TB memory in order to open the near infinite computing.
2) Operating System
Based on years of technology accumulation, Alibaba has built a powerful operating system array consisting of its own servers, terminals, and IoT operating systems.
Server Operating System
The Alibaba server operating systems include the single server operating system and the Alibaba Cloud Linux suitable for the cloud environment. Alibaba has insisted more than 10 years of independent research and development and innovation. Alibaba Cloud Linux has the characteristics of full stack optimization, open source, security and reliability. Alibaba Cloud Linux has been open source since its release in 2015, and Alibaba Cloud Linux 2 was released in 2019, attracting the co-construction of large enterprises at home and abroad, such as Intel and Toutiao.
On-Terminal Operating System
Alibaba developed the on-terminal operating system (AliOS) for the interconnection of everything. This system has been applied to multiple fields, such as smart phones, tablets, televisions, and Internet vehicles. Currently, Alibaba has established deep cooperation relations with self-owned brands such as Roewe of Shanghai Automotive Industry Corporation and Morris Garages as well as joint ventures such as Ford and Dongfeng Citroen. The sales of vehicles equipped with AliOS exceed 1 million.
IoT Operating System
Alibaba has developed a microkernel-based IoT operating system, AliOS Things. It has supported more than 100 sensors and more than 100 chips, with a cumulative shipment of more than 100 million. It is widely used in smart home appliances, intelligent hardwares, industrial control, medical equipment and other scenarios.
Alibaba’s database technology has passed the process of “business-open source-self-developed”, continuously breaking through database technology and application ecology, and realizing the autonomous control in the database field. Since 2010, Alibaba has been developing its own databases. It has seized the opportunity of technological transformation in cloud databases and developed a series of advanced database products, joining the world’s leading camp.
Alibaba’s database products not only support the largest e-commerce environment in the world with high-concurrency and low-latency. The database products provide the real-time data services for the online business data of tens of millions of small and medium-sized enterprises. The database products also provide the stable and reliable data storage and analysis services for government affairs, manufacturing, finance, communications, customs, transportation and other key industries. The current enterprise users exceed 100,000, including the State Administration of Taxation, China Post, Sinopec and China Unicom.
Alibaba Group was included in Gartner’s Magic Quadrant for global databases for two consecutive years and was placed in the Challengers quadrant in 2019. Alibaba is the unique selected Chinese science and technology enterprise.
Technology autonomy and controllability: Alibaba has carried out independent research and development in an all-round way to overcome key technologies such as separation of storage and computing, and has obtained nearly 80 relevant patent licenses. At the same time, Alibaba has carried out the limit optimization for domestic chips and operating systems, to reduce the research and development cycle for different domestic chips.
Leading the cloud-native era: Based on the concept of cloud native, Alibaba widely uses the computing and storage separation technology to make the computing layer serverless, and implements the flexible and elastic architecture.
Intelligence: Alibaba’s intelligent database technologies include two core components: intelligent kernel and intelligent O & M and control. Alibaba has already achieved the highest level of technology in the world.
After more than 10 years’ painstaking research and development, Alibaba already has the most powerful and rich full-stack database product series in China, including relational databases, non-relational databases, analytical databases, and migration tools, about 30 products. These products can meet different database application development requirements and help enterprises migrate their core business to the cloud.
Intelligent control: Alibaba has developed the intelligent database management and control platform to implement intelligent diagnosis, intelligent recovery, and intelligent optimization. The platform is used to manage large-scale database clusters and provide stable, efficient, and easy-to-use database services.
Security credibility: Alibaba integrates with trusted hardware, blockchain, authenticated data, and encrypted computing technologies to protect user data in an all-round way, including data storage and computing encryption, query verification, data privacy protection, and data tampering detection.
The database R & D team is led by Li Feifei, a leading figure in the global database field, an outstanding scientist of the American Association of Computer Science (ACM), and a lifetime professor at the Department of Computer Science at the University of Utah in the United States. The team members have reached more than 700, including many core research and development personnels from Technical Comittee on Databases and Task Force on Big Data, China Computer Society (CCF). More than 95% of the core research and development personnels graduated from top universities at home and abroad. The team’s papers have been selected on Very Large Data Base (VLDB), Special Interest Group on Management of Data (SIGMOD), International Conference on Data Engineering (ICDE), the top three academic conferences in the database field.
4) Big Data Computing Platforms
The big data computing platforms of Alibaba have served more than 5,000 large public cloud customers and 200 private cloud customers. The number of servers for offline computing exceeds 100,000 and remains the top one in TPCx-BigBench, achieving the largest data scale, the highest processing request per minute, and the lowest unit cost in the world. During the Double 11, Alibaba Cloud is able to process more than 600PB of data during that time, supporting over 70% of the business growth with an annual hardware growth rate of 20%.
In addition, Alibaba acquired Data Artisans, the parent company of the well-known open source stream processing framework, Flink. Meanwhile, Alibaba’s big data engine Blink is officially open source as a branch of Flink.
5) Internet-based Dual Mid-End
The traditional IT architectures focus on process automation. With the continuous development of enterprises, large and medium-sized enterprises generally have dozens or even hundreds of IT systems, with issues that each system is independent, the service functions are repeatedly constructed and data cannot be collaborated. After enterprises move from informatization to digitization, the “mid-end strategy” will become the core that supports the rapid development and innovation of enterprises.
The dual mid-end architecture (business mid-end and data mid-end) of Alibaba is in line with the evolution trend of enterprise IT architectures in the era of Internet big data. It breaks the chimney-like architectures between different business departments, breaks the data islands, and realizes the goal of “all business becoming data”.
Alibaba’s dual mid-end has helped Sinopec build the EPEC platform in 90 days, supporting transaction volume of RMB 227.6 billion. It has helped Haidilao establish an intelligent service system with 30 million members within 5 months. This system not only makes basic functions such as scheduling, reservation, and ordering smoother, but also creatively integrates new technologies such as communities, short video sharing, and intelligent voice interaction. Through dynamic expansion, it can support marketing activities of hundreds of millions of members and tens of millions of participants.
Development of DAMO Academy
The Alibaba DAMO Academy for Discovery, Adventure, Momentum and Outlook has set up research areas such as machine intelligence, data computing, robotics, financial technology and X, namely the “4 + X” research field, and established 14 laboratories. Since its inception, DAMO Academy has widely recruited top scientists and researchers around the world to quickly form the wild goose queue effect. A group of outstanding engineers have been selected to cooperate with the transformation and implementation of innovative technologies to practical applications. At the same time, a group of excellent doctoral students from the world’s top institutions of higher education have been recruited, and a new scientific research force has been reserved to form the reasonable “head-waist-leg” personnel structure. At the same time, it attracts high-end talents and seizes the talent highland by using various flexible ways such as consultants, visiting scholars and post-doctoral workstations.
DAMO Academy also conducts multi-disciplinary and multi-field academic cooperation with top universities and scientific research institutions in the world, to build a global industry-university-research cooperation ecosystem, and to explore a talent resource model that integrates autonomous research, cooperation and openness. It has established joint laboratories or research centers in world-famous colleges and universities. In scientific research projects, it cooperates with more than 100 scientific research teams from over 50 well-known colleges in more than 10 countries around the world, and they have jointly trained more than 100 graduate students, achieving a number of excellent academic achievements.
1) Artificial Intelligence
Alibaba DAMO Academy has built a comprehensive machine intelligence algorithm system, which covers fields such as speech intelligence, language technology, machine vision, City Brain, and decision-making intelligence. It has made multiple world-leading achievements.
It has exceeded the world record in speech recognition and dialogue. In terms of speech synthesis, the self-developed Knowledge-Aware Neural TTS (KAN-TTS), a speech cooperation technology. This technology can improve the naturalness of specific speaker data to over 97%, which is expected to pass the Turing test.
In 2018, it won five championships in Machine Translation competition WMT (Workshop on Machine Translation), the internationally recognized top machine transaltion competition. In January 2018, it refreshed its ranking in the Stanford Question Answering Dataset (SQuAD), the top competition in the field of machine reading comprehension launched at Stanford University. This is the first time that the artificial intelligence ability of reading has outperformed the humans’. At present, the machine translation technology team of DAMO Academy has implemented 48 language translation directions, with daily calls reaching 0.75 billion times to provide online multi-language services for cross-border e-commerce in countries along “The Belt and Road” route.
In January 2018, it was the champion of three road scene segmentation tasks in KITTI, the world’s largest automatic driving computer vision algorithm set. In November 2018, it won two championships in DeepGlobe, the global remote sensing image intelligent analysis competition, reaching the world’s highest level in the field of remote sensing image road network identification and land use identification. In 2019, Alibaba published 17 papers at the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), a top international conference in the field of computer vision and pattern recognition, which represents the most advanced level of the development of machine vision in academia world and industry world.
2) Chip Technology
Alibaba founded the Pingtouge (T-head) semiconductor company in 2018. This company has focused on the research and development of embedded CPU and AI chips to create a secure, intelligent and inclusive cloud-based integrated computing support.
Hangzhou C-Sky Microsystems Company, acquired by T-head, has been deeply engaged in the field of embedded CPU for 18 years and has developed a fully autonomous and controllable embedded CPU. It is the only company in China that realizes large-scale mass production. At present, more than 200 CPU IP licenses have been achieved, and the cumulative mass production has exceeded 1 billion. This series of embedded CPU is also a landmark achievement of domestic CPU supported by the National Science and Technology Major Project-Core Electronic Devices, High-end Generic Chips and Basic Software.
Embedded SoC Common Platform
In view of the fragmentation difficulties in the IoT chip industry, the company has carried out research and development of common system chip platforms and launched the common SoC platform. The platform greatly reduces the research and development difficulty of embedded chips in specific fields. Through the foreign cooperation and authorization, the platform helps the industry to reduce chip design costs and thresholds.
Artificial Intelligence Chips
T-head’s product plans for AI chips range from dedicated inference chips to universal training/inference chips, to IoT AI chips integrated with embedded CPUs, which forms a smart computing system integrating cloud, edge, and end. Currently, the first generation of AI chips have been successfully applied to cloud services, City Bain, Industrial Brain, and other businesses.
3) Quantum Computing
DAMO Academy’s quantum laboratory is committed to unlocking quantum potential and exploring a new disruptive computing power in the future. The quantum laboratory has set a goal of reaching international leadership in the two core directions of high precision and design automation, and fully competes with international teams on the same stage.
In April 2019, the “Lab-1”, a hardware laboratory of quantum laboratory, was built in Hangzhou, with a construction area of about 600 square meters. It consists of the low temperature measurement and control laboratory, the preparation laboratory and various auxiliary equipment and systems.
The quantum computing lab of Alibaba DAMO Academy has nearly 30 full-time employees, and all of them come from prestigious universities and enterprises at home and abroad.More than 90% of the members have doctoral degrees. More than 85% of the members have overseas study experience. The team leader is Shi Yaoyun, a former tenured professor at the University of Michigan.
4) Automated Driving
The automatic driving research team of DAMO Academy is committed to applying the automatic driving technology to the China Smart Logistic Network. This team has developed the “Xiaomanlv”, the logistics robot, by using the L4 automatic driving technology. This robot integrates the cutting-edge artificial intelligence and self-driving technologies of DAMO Academy and has the human-like cognitive intelligence. Its autonomous driving rate reached 99.9999%. It utilizes the state-of-the-art AI and autonomous driving technologies, and provides the industry-leading competitiveness in terms of intelligence, security, and mass production.
National Open Innovation Platform for Next Generation Artificial Intelligence on Intelligent Vision
Construction of National Science and Technology Innovation System
Based on national strategic needs, cutting-edge technology research, and major social and economic propositions, Alibaba deeply participates in the construction of the national science and technology innovation system. Alibaba has undertaken more than 50 national and local major scientific research and industrialization projects, and is applying for more than 40 projects. In 2017, the City Brain of Alibaba was included in the first batch of National Open Innovation Platform for Next Generation Artificial Intelligence on Intelligent Vision. For city level massive data, Alibaba has independently developed the core AI computing engines such as functional online learning, heterogeneous reasoning, model acceleration, graphic fusion and city map search. By the City Brain, the query of tens of billions of data in a city can be completed within seconds, and the video computing speed can be increased by 8 to 10 times.
The innovation platform is committed to establishing the technology and application ecology of the urban intelligent computing platform, and carries out technical cooperation with more than 10 university research institutes and dozens of enterprises.
The innovation platform undertaking team has developed a large-scale visual computing platform (Tianqing) of city brain, a traffic prediction and Intervention (Tianyao), a progressive image search engine (Tianying), a 3D visual computing and reasoning (Tianpu), a traffic prediction and intervention (Tianji) and an urban comprehensive management and identification (Tianjing). These products have been widely used in more than 40 cities and more than 60 cities at home and abroad, and achieved positive results.
“The prosperity of science and technology leads to the prosperity of the nation, and the strength of science and technology leads to the strength of the country.” Today’s world is experiencing a great change that cannot be seen in a hundred years. Economic development and people’s livelihood cannot be separated from scientific and technological innovation. Respecting for scientific and technological talents, adhering to independent innovation, and having the courage to sit on the bench and come up with real original results are the core strategies of national development, which Alibaba has always adhered to.
Alibaba will continue to work tirelessly and forge ahead with all the enthusiasm and optimism.
The views expressed herein are for reference only and don’t necessarily represent the official views of Alibaba Cloud.