Current location - Loan Platform Complete Network - Big data management - Big data cluster wave
Big data cluster wave
In 2020, IDC Global Internet Data Conference was officially held in Shenzhen. From the perspective of data economy, the conference interpreted the mainstream technology trends such as new infrastructure, edge computing, container technology, AI and RPA. Many guests and enterprises shared their views and thoughts on the platform and technology ecological construction.

Chen Yicong, chief technology officer of Inspur Group Guangdong Company, attended the conference and shared his views on smart computing and data centers with the theme of "Breaking through computing boundaries and opening up to achieve the future".

Chen Yicong said that both Internet services, data services and data center operations are closely related to computing. With the advent of the era of 5G, AI and big data, the definition of computing is no longer limited to the traditional decentralized computing, but linked to wisdom. At the same time, the new business requires more and more computing power and data center, so it is not limited to the original definition. Breaking through the computing boundary and promoting the formation of an open computing ecosystem have become the core points to be solved urgently in the industry.

He emphasized that intelligent computing is constantly changing our current and future production and lifestyle, leaving a profound impact from daily life to digital transformation of enterprises. In short, computing power is productivity.

Chen Yicong said that the traditional data center can meet the computing needs by deploying a computer room or a cabinet of equipment. However, with the development of Internet business, the expansion of data and the outbreak of epidemic in early 2020, many offline activities are held online. The development of online business means that the scale of data center is getting larger and larger, and more and more devices are carried, which also means that new requirements are put forward for the technical support and computing power of the back end.

In addition, this year, the National Development and Reform Commission put forward a plan to lay out about 10 regional data center clusters and intelligent computing centers throughout the country, and Guangdong-Hong Kong-Macao Greater Bay Area has also planned data center clusters, so the construction of new data centers has become the focus of attention. From the expansion of data center scale a few years ago to the increase in the number of computing equipment in super-large data centers, the construction of super-large data centers has been growing rapidly, and the development of large data centers has brought the complexity of application support and data center management.

Chen Yicong said that the original equipment such as servers, storage, networks and computer operating systems needed for data center construction is not a problem that needs to pay too much attention to its manufacturers and equipment brands. However, with the increasing gathering of big data, the recognition of cloud unification has attracted more and more attention and thinking from the industry. Whether the unified cloud can manage all servers in China, including Alibaba Cloud and Tencent Cloud, remains to be discussed.

The biggest expense of data center is electricity. At present, the total energy consumption data of all data centers in the world shows 300 billion kWh a year, which is equivalent to the annual power generation of 30 large nuclear power plants. It is unrealistic to build a super-large data center cluster in Shenzhen to serve the whole of Guangdong-Hong Kong-Macao Greater Bay Area. At the same time, it is necessary to build three nuclear power plants for power support, which will inevitably put forward higher requirements for the green, energy-saving and efficient operation and maintenance of the newly-built super-large data center.

So how to solve this problem? Chen Yicong gave an example. He said that the original server was a physical machine, and now cloud computing, which is gradually serving, has been accepted by the industry. The cloud can be used immediately, but there are also some performance losses; Physical machines can be delivered in a few hours at the earliest. All physical machines are turned on in real time, that is, they are ready to use, with strong isolation and stability, but compared with cloud computing, they also consume the most energy.

On the other hand, the ultra-high energy consumption has seriously affected the rapid development of data center business. Because of the rapid development of Internet in China, our data center construction is also in the forefront of the world. The annual electricity consumption next year is 1000 billion kwh, which is equivalent to the annual power generation of the entire Three Gorges Dam. Energy consumption control puts forward requirements for some new technologies. At the same time, breaking the physical boundary and realizing resource pooling is also the development direction of data center, in order to improve the utilization rate of computing resources and achieve higher performance, lower TCO and efficient operation and maintenance.

Inspur delivers goods through L 1 1 level integration, which makes delivery faster and more convenient. Three weeks before the Spring Festival on 20 19, the supply guarantee of 10000 sets with sudden demand was completed, and it took only two weeks from receiving the demand to delivery. It has created a delivery record of 1 0,000 servers to the data center within 8 hours every day, ensuring the huge business demand of customers to realize the global audience red envelope interaction of CCTV Spring Festival Evening. In addition, Inspur has also accelerated the construction of intelligent computing centers through open computing and promoted the transformation of social wisdom. At present, the whole line of open computing standard product lines has been laid out, and the concept of open computing has been practiced for many years, leading the open computing standards and continuously defining leading open computing products.

Chen Yicong said that Inspur has also been recognized in the market. Not long ago, the Q2 Data Report released by IDC, an international authoritative data organization, showed that domestic shipments of Inspur servers accounted for 1 1.3%, with an increase rate of 80%, ranking first in the world. The market share of Inspur 86 server in China is 42.6%, ranking first in China. Especially in the field of AI, the market share of Inspur AI servers in China has exceeded 50% for three consecutive years.

He emphasized that Inspur has provided a full-stack information solution for the industry, and also looks forward to working with industry partners to create an open computing environment and create a better future for the industry's ecological construction.