The “danger” and “machine” in the operation of data center

In the past two years, “new infrastructure construction” has become a hot word in the data center industry. The national development and Reform Commission has clearly pointed out that the “new infrastructure” mainly includes information infrastructure, integrated infrastructure and integrated innovation infrastructure. Among them, information infrastructure is the most important part of ICT industry, including communication network infrastructure represented by 5g, Internet of things, Satellite Internet, etc., new technology infrastructure represented by artificial intelligence, cloud computing, blockchain and other technologies, and computing infrastructure represented by intelligent data center and Intelligent Computing Center.

with clear guidelines and clear investment plan list of key projects of local governments, we can clearly see the importance of data center industry development under the wave of “new infrastructure”. From the perspective of industry analysis data, since the introduction of policy guidance, the data center industry has once again set off a climax. It is a great challenge and a new opportunity for us “data center people” to help traditional industries upgrade and transform to high-tech goals guided by the government.

since the introduction of the new national infrastructure policy, the statistical analysis of regional distribution shows that the planned total number of new cabinets in East China, North China and South China has all impacted 500000, and the north, Shanghai, Guangzhou, Shenzhen and surrounding areas continue to lead the country. Other provinces and cities have also seen an investment boom, and the construction volume of data centers in China has increased dramatically. As of July 2020, the data center with more than 2000 cabinets can be defined as “super large data center” by the end of 2020. The planned total number of new single projects is about 300, and the planned total number of new cabinets is about 2.2 million.

data center is the bottom layer of infrastructure, the carrier of all information technology and data application, the hub of information communication, and the precondition of building information platform. As a data center developer and operator, Qingyun qingcloud is duty bound to empower the construction and operation of data center with scale and quality. But the real higher requirement is that we complete the digital revolution in the data center industry, provide intelligent operation solutions, deliver core infrastructure that can keep pace with the pace of “new infrastructure”, and provide strong support from the bottom to the takeoff of digital economy.

an epidemic situation shows us that China’s institutional advantages are the fundamental guarantee to overcome the epidemic situation. It is hard to imagine that huoshenshan hospital, which shocked the world under the guidance of the state, has sprung up; it is hard to imagine that automobile manufacturers can change into the production center of medical and health products such as masks in an instant.

similarly, under the “new infrastructure” under the guidance of the state, we have seen steel enterprises such as Hangang, Jigang and WISCO ready to move at the door of the data center. These “new players” have policies, land, and water and electricity. They have already occupied all the inherent advantages from the beginning, and will have a bright future in the data center market.

in recent data center construction projects with state-owned assets background, owners are no longer satisfied with the intelligent system of traditional data center, and even some owners begin to reject the traditional construction mode. In a large data center project recently launched, the owner’s requirements for the intelligent degree of self built data are beyond imagination. Robot inspection, UAV inspection, holographic sand table speech, Mr maintenance assistance and other scientific and technological means are required to be realized. I believe that in the next few years, the list of China’s IDC industry’s top 100 will be comprehensively shuffled, and the intelligence of data centers will certainly affect the fate of operators.

has friends in the circle in recent days. “Data center sales are not good, operation cost is difficult to optimize, the existing intelligent system is still in use, but the manufacturers that have maintained it have been out of use, and have been using it for a long time, but many systems make complaints about it. If the replacement cost is too high, it is really tasteless, and it is a pity to abandon it.”

in fact, it’s because it’s not intelligent. If the system platform we provide is not a product, but a “IDC app store”, all the intelligent functions are defined in the cloud software, and the AI algorithm is introduced to issue or upgrade at one click, except for the “hard demand” that meets the intelligent requirements of the data center itself. At the same time, the system algorithm can also be continuously optimized and upgraded, providing new value-added business function scenarios at any time, realizing the “pluggable” function of the function scenario module, and realizing the decoupling of software and hardware, refusing to be “kidnapped” by hardware conditions.

in fact, these words have not been fully explained. In this era of 5g, equipment manufacturers and ecological partners have been supporting Qingyun qingcloud intelligent platform to develop 5g terminal. In the current environment, what we are most afraid of is imagination. If our intelligent terminal hardware is also “pluggable”, then the life cycle of this system is theoretically “infinite”? In this area, Qingyun qingcloud has actually taken a big step. Please remember that “sweat operation and maintenance” has been unable to catch up with the pace of the “new infrastructure” era. The so-called traditional intelligence, which is strongly bound to hardware, is difficult to iterate and upgrade, and the technology is eliminated quickly. There is no need to copy and promote it.

this year, with the outbreak of “Xinguan”, the traditional data center operation mode has been unable to meet the needs of end users. Data center operation to reduce the dependence on people, reduce the direct contact between people is imminent. Similarly, in the state-owned new data center project, we also see a very strong demand for intelligence and digitization. The data-based operation and management system of the data center provided by US reduces the manual content by 80%, and the intelligent system greatly reduces the personnel skill requirements.

there is a joke: “there is a power failure liability before, and a customer claims chopper later”. Qingyun qingcloud reduces the manual inspection content through intelligent AI inspection. With the support of AI algorithm, we transform the “passive inspection” into “active identification”, and combine with the “no dead corner” visual solution to do a good job of high-risk risk prediction in advance, so as to prevent accidents in the bud.

this is an old topic, power usage effectiveness, energy efficiency index of data center. In the current market environment is so transparent, how to reduce their own costs is also the key to win. Traditional refrigeration system mode optimization ability is limited, generally reduce pue & lt; 5%. Usually, in order to reduce pue, operators will pay a considerable amount of hardware transformation investment and recover the cost through long-term operation. Through software, Qingyun qingcloud AI technology can reduce pue by 5% ~ 15% with the lowest cost input and the fastest cost recovery. Only in this way can we achieve the ultimate goal of reducing the overall cost and strengthening the market competitiveness.

for the optimization of traditional pue, we rely too much on building automation system. At present, there are more than 3000 enterprises doing relevant systems in China. At present, Honeywell, Johnson and Siemens account for half of BA system, and the other half is divided by other brands, mainly including Invensys, Haiwan, Tsinghua Tongfang and ASI. The three brands have the most complete product series, with a full range of products and high market evaluation; among other brands, Gulf supports lon agreement, which is suitable for places with small system capacity and few points, with unclear hierarchical structure, unstable system capacity, and only less than 1% market share. ASI only has sensors, valves and other components are mostly OEM. Due to the strong sensors, they are widely used in places that need accurate control, such as electronic factories. Tsinghua Tongfang’s rh2000 series is self-developed and has its own intellectual property rights, but its power consumption is twice that of foreign products, so it lacks competitiveness. There are few enterprises with real R & D strength. Most of them maintain the sales of related hardware, and there is a considerable gap between them and foreign enterprises. Most of the systems are flashy. Traditional Ba manufacturers can not bring real convenience and energy-saving effect to customers from the aspects of communication, scheme, connection, installation and maintenance. According to statistics, the opening rate of BA system is less than 60%, and the satisfaction rate of using is less than 30%. At present, only 3% of the domestic traditional Ba systems can work effectively and achieve the expected results.

after years of experience in data center operation, the cooling link parameters of the data center are many, and each adjustment needs an observation period, and all of them are completed for at least half a day, so the optimization effect is poor; Qingyun qingcloud energy efficiency optimization technology is based on AI algorithm, accurately calculates the accurate expression function factor of pue, calculates the optimal control factor combination, and all parameters are once Data center energy saving is a system engineering, more importantly, the key physical infrastructure as the support core of the data center, to ensure the safe and reliable operation of the information system is its first priority, and energy saving can not be at the expense of the security and reliability of the data center. The system parameter reading and instruction sending of Qingyun qingcloud are carried out through the original group control system, and the terminal equipment is not directly read or controlled; the system and the original group control system are independent, and the shutdown does not affect the operation of the original group control system, so it can complete the task safely and independently.

in this era, don’t believe in the “unattended” Data Center for the time being. At least in this era, it can’t be completely “unmanned”, but we should minimize the use of personnel. Personnel management system is still a very important part of data center operation. Therefore, in terms of platform design, the personnel management system should also be established in strict accordance with the requirements of uptime M & amp; o. From the continuous optimization of post JD, training plan, skill certification, echelon construction and process manual of operation and maintenance personnel, it comprehensively covers the practical ability of effective operation and maintenance.

the smart data center of Qingyun qingcloud is designed with “one standard, two platforms and three plates”, and takes “unified certification, unified management, intelligent linkage and process specification” as the standard, and divides the “end-to-end linkage”, “integrated operation and maintenance” and “visual components” around the two platforms of “intelligent operation and maintenance platform” and “Internet of things platform” The landing of a variety of integrated scenes. The overall structure of the scheme is as follows:

technical details are not discussed here. At present, we think that intelligent upgrading and digital transformation of data center have become important objectives of industry development. Integrating the whole industry service chain, building industry service platform, and realizing the integration with Internet, Internet of things and other information technology have become important development means. Data center operators can’t wait for the “revolution” of intelligent new forces. Through intelligent operation, algorithm optimization and upgrading of data center must be a long-term operation practice process, not only a hardware purchase and sale, but also a new operation methodology and the introduction of long-term iterative upgrading optimization system in the future. Experts in the data center industry, do you agree with the above development direction? If the answer is yes, then you and we have the same ambition and goal. If we have a short way to cloud, we hope that we can shorten the path of cloud transformation.

to sum up, the number of

Author: zmhuaxia