What kind of data centers are we need in the vision of "carbon neutral"?

With the continuous advancement of the "new infrastructure" process, the progress of digital transformation and upgrading has accelerated, and data centers, as the strategic resources and digital infrastructure of future economic and social development, are inherent in explosive growth, their computing power is uneven, high energy consumption is high, and energy consumption is high. The problems such as insufficient management of refined management, and the high -quality development strategy of the data center has become an opportunity for the industry to deepen digital transformation in the industry.

Regarding the high energy consumption of data centers, the annual values ​​given by different domestic research institutions are different. Some research institutions predict that in the next ten years, the power consumption of domestic data centers will double. In fact, with the application of new energy -saving technologies of data centers, although the demand for computing power continues to increase high proportion, the situation of high energy consumption and high carbon emissions in the future data center may not reach such a high level, but this does not mean The IT industry and decision makers can let go of it. The development of green energy -saving and low -carbon in the data center is still a key topic worthy of attention.

In September 2020, my country clarified the goal of "carbon peak, carbon neutrality", marking China's determination to promote high -quality economic development, social prosperity and ecological environmental protection. In February 2021, the State Council issued the "Guiding Opinions on Accelerating the establishment and improvement of the Economic System of Green Low -Carbon Cycling Development Development Economic System", requiring to accelerate the green transformation of the information service industry, do a good job of large and medium -sized data centers, green construction and transformation of network machine rooms, establish green operations and maintenance system. In June 2021, the "Fourteenth Five -Year Public Institution Plan for Public Institutions" jointly released by the State Administration of Management and the National Development and Reform Commission and the "National Development and Reform Commission, the Central Cyber ​​Information Office, the Ministry of Industry and Information Technology, and the National Energy Administration's four ministries and commissions" Implementation of carbon -peak carbon neutralization and target requirements to promote new infrastructure such as data centers and 5G green and high -quality development implementation plan "requires large and large data centers to operate electrical energy utilization efficiency to drop below 1.3. In July 2021, the Ministry of Industry and Information Technology released the "Three-year Action Plan for the Development of the New Data Center (2021-2023)", which proposed to adhere to the concept of green development, support the application of green technology, green products, and clean energy, and comprehensively improve the energy utilization efficiency of new data centers. Essence

With the successive promulgation of a series of national policies, the development of data center green and low -carbon development has emerged as an important part of the high -quality development strategy of data centers. The development of data centers to enhance the energy efficiency of equipment through innovative energy -saving technology, make full use of renewable energy to reduce carbon emissions, and then promote data centers as the basic supporting role of digital new infrastructure, and drive the society to achieve the goal of energy saving and carbon reduction.

01

What is the "green" data center?

Green data center aims to maximize energy efficiency and minimize environmental impact. Generally speaking, including the performance efficiency, environmental impact, resource integration and energy coordination of data centers.

Performance efficiency is the basic requirement, and high -reliability data centers need to meet the requirements of the A -level data center in GB50174 "Design Specifications for Data Center". On the one hand, it is necessary to maintain the data center without failure, and on the other hand, the IT system, refrigeration, lighting and electrical systems of the data center can maintain stable and efficient operation. Environmental influence mainly indicates the impact of the external environment during normal production and operation. During the construction and construction stage, the use function and planning positioning of the building should be combined with the comprehensive environmental impact assessment of the projects including the comprehensive environmental impact of atmosphere, water, soil, sound, solid waste and ecology. Resource integration mainly examines the material selection, construction technology, equipment or material recycling and renewable utilization rate and product packaging in the construction of data centers. Energy coordinating focuses on energy management and energy -saving measures after data center operation.

At this stage, the evaluation of the green data center is mainly to consider its performance efficiency, take into account the energy coordination of the data center operation, and improve the overall energy use efficiency.

Data centers are also very important because of the "three major problems" of high heat heat dissipation bottlenecks, resource utilization and low energy efficiency, and their green energy saving is also very important. From the country to the local level, the energy consumption of data centers has been guided and restricted.

Figure 1: Energy use of data centers

The main energy consumption of data centers is IT device, power distribution system and HVAC system, as shown in Figure 1. A typical data center, the energy consumption is mainly composed of four major blocks: the largest block, which is the IT device system that accounts for about 50%of the total energy consumption of the data center, including server equipment, storage equipment and network communication equipment; second Large pieces are air -conditioning systems that account for about 38%of the total energy consumption of data centers. Among them, the air -conditioning refrigeration system accounts for about 25%of the total power consumption, and the air conditioner delivery and return air system account for about 13%of the total power consumption; the third largest block; the third largest block It is the UPS power supply system that accounts for about 10%of the total energy consumption of the data center. Among them, the UPS power supply system accounts for about 5%of the total power consumption, and the UPS input power supply system accounts for about 1%of the total power consumption; 1%of the remaining power consumption of the center belongs to the auxiliary lighting system.

It can be seen from the above analysis that the IT device of the data center is the maximum energy consumption. The largest energy consumption is the computing device represented by the server, so it is necessary to focus on the server energy efficiency. The energy consumption of HVAC system in the data center accounts for about 37%of the total energy consumption of the data center. This part is also regarded as the potential part of the energy center energy saving. The energy consumption of the power supply system accounts for about 10%of the total energy consumption of data centers, mainly from the conversion of transformers, UPS and other power supply distribution systems, followed by lighting. Therefore, the green energy saving of the data center must make efforts from the computing devices, such as server, HVAC and electrical systems. 02

How to improve the efficiency of data center energy use?

The improvement of the overall energy efficiency of the data center is inseparable from the energy efficiency improvement of each composition equipment, or the energy efficiency of the equipment can directly effectively improve the energy efficiency of the data center, such as improving the energy efficiency of the server and reducing power supply loss. However To the bottleneck, to a certain extent due to the limitations of technical and cost, its energy efficiency cannot improve or increase energy efficiency to increase sharply. After encountering difficulties in equipment energy efficiency, it is usually adopted to improve the system energy efficiency. For example Matching, because the elasticity of the system energy efficiency is relatively large, usually starting from small systems, gradually introducing more variables, more and more systems involved, and then considering management factors, it has become the method and method and method of improving energy efficiency from the organizational level. way.

03

Energy -saving technology

3.1 server

The essential role of the data center is to provide computing power resources, and the largest energy consumption proportion of data centers is the energy consumption of information equipment, aside from computing power. It is unreasonable to simply pursue extremely low power use efficiency (PUE). Many semiconductor manufacturers such as Intel, Samsung, Gaotong, Nvidia have made great efforts, and all kinds of accelerated chips are blooming. XPU (CPU, GPU, DPU, etc.) has become a new track for semiconductor chip manufacturers to compete, and a clear main competition line is gradually obvious -major chip companies are building their own diversified product capabilities. Among them, Intel, based on the XPU product strategy, created a variety of heterogeneous resources from CPUs to GPU, FPGA, IPU, etc., with XPU+OneAPI as a grasp of heterogeneous computing, creating a comprehensive product portfolio from cloud to end, covering the CPU , GPU, IPU, FPGA, and special ASICS solutions, solve the different performances of different computing scenarios caused by different architectures and different instruction sets, better cope with the diversification of calculations, and provide provision Strong computing power and sufficient flexibility, and have advantages in terms of power consumption, reliability, volume, etc., solve the high cost and higher incidence of calories brought by more processors when processing massive data processing. question.

The increase in the computing power of the server chip, and the diversification of the calculation scene is essential for data center energy saving. Intel has greatly improved the load processing speed through dedicated acceleration chips, that is, the same workload is fast. GPU and AI, provide higher calculation density and faster calculation speed. Among them, Intel Data Center GPU, which is about to be launched soon, is an ARCTIC SOUND-M (ATS-M), can provide 150 trillion computing (150 TOPS) per second. The first flagship data center GPU -PONTE VECCCHIO has shown an excellent advantage in complex financial service applications and AI reasoning and training workloads. In May this year, the Habana GAUDI2 processor dedicated to high -performance deep learning AI training, as well as the PONTE VECCHIO GPU based on the XE HPC micro -architecture and designed for high -performance computing and AI can allow end users to support the diverse architecture and allow end users Make full use of the high performance and high energy efficiency of the processor.

In the energy consumption structure of the data center, the energy consumption of the refrigeration and air -conditioning system is second only to the energy consumption of information equipment. How to reduce the cooling energy consumption of the data center has become the key to reducing PUE. With the increase in the power consumption of server units, the original server cabinet of the original size can accommodate the server power of the server often exceed 15kW. In the case of the existing air -cooled data center, this has reached the bottleneck of the air to the flow heat dissipation capacity. As a technology with stronger heat dissipation capabilities, liquid cooling technology can help higher power density. The liquid cooling technology refers to the cooling method of using high -ratio heat capacity and high heat -up coefficient as the working quality of the heating transmission to meet the heat dissipation requirements of IT devices such as the server. In other words, the air is used to replace the air through liquid to take away the heat generated by the CPU, memory barrier, chipset, expansion card and other devices during runtime. Common liquid cold technology includes three main forms: cold plate, immersion and spray type. The liquid cold server can accept higher cold sources for returning water temperature, maximize the outdoor natural cold source, effectively reduce the data center PUE to Within 1.2. Lenovo is based on the new generation of sea god warm water and cold technology and the third -generation Intel® Xeon ® expansion processor, creating Lenovo ThinkSystem SD650 server, which uses innovative cooling technology to improve performance and reduce power consumption. At Shanghai Jiaotong University, the "Siyuan No. 1" high -performance computing cluster was created, and PUE could be as low as about 1.1, achieving 42%energy conservation and emission reduction. Although the liquid cooling server has greatly improved the heat dissipation efficiency, due to the poor standardization level, the industry has not yet had a unified design specification, so reliability still needs to be improved, and the cost needs to be reduced. Intel Juli joined hands with industrial ecological partners to open innovation, build liquid cooling solutions, research in four horizontal directions of chips, servers, racks, and data centers, explore more economical and efficient cooling solutions for design reference, and is committed to proposing cold boards The requirements of the design of the liquid cooling system, and the specification requirements of the future liquid cooling design, provide paths and reference for the design and research of the data center liquid cooling scheme. Its standardization, reducing design and use costs, thereby promoting the establishment of and improving the cooling of cold plate liquid cold ecosystems to promote the maturity of the entire industry.

3.2 Cabinet -level energy -saving optimization technology

Cabinet -level energy -saving technologies mainly include: load changes caused by business volume, business type, etc., automatically adjust the power; use a comprehensive power supply method to improve the power density of the single cabinet; increase the calculation density, unify power supply and share heat dissipation management management Wait for technical measures.

According to Intel's measured data, the use of the rack backup battery to eliminate the external peak power consumption, which can increase the server's listing rate by 20%to 30%. Increasing the calculation density, by increasing the bus voltage and adapting to the needs of high -power racks, the power supply efficiency can be increased by 2%, and the space utilization rate of the cabinet can be comprehensively improved. The modular design implements the concept of green energy -saving, and realizes unified power supply and sharing heat dissipation management. Liquid cooling and cold plate heat dissipation, with professional refrigerant, covering main components such as CPU, memory, the whole machine PUE as low as 1.1, and can greatly reduce the overall ownership cost. The processor power control is performed, and the power is automatically adjusted according to the business load, and the micro -code based on Intel -based open processors, adjust the motherboard and processor voltage. Practice has shown that this solution has significantly increased power density, which can support the power density of up to 20kW, and achieve the effective improvement of computing power. Even if pure wind is cold, PUE can maintain internal control below 1.2.

In addition, Intel joined hands with China Telecom to promote the deployment of AI energy -saving technology. The energy -saving rate of the refrigeration system exceeded 23%, and the average PUE PUE decreased from 1.49 to 1.38; Make the refrigeration PUE below 1.1, and the stand -alone cabinet supports 144 high -power CPUs.

04

System energy efficiency improvement technology

A reasonable end air -conditioning airflow is the basis and premise of energy saving and consumption reduction of air conditioning systems. Improving the return of precision air -conditioning back air temperature and the supply of cold sources have always been the consensus of the energy saving of the data center HVAC, but often the above -mentioned energy -saving measures are not achieved due to the unreasonable ending airflow. The reasonable airflow tissue is reflected in the air -conditioning area of ​​the cabinet in the data center machine room uniformly distributed, and the temperature field of vertical and horizontal direction is in a relatively balanced state. You can use related air volume, wind speed, and temperature field test instruments to carry out measurement and evaluation of airflow tissue to parameterize and specific airflow fields. Through the application of measurement technology, explore the coupling association between the airflow tissue and the temperature field distribution of the data center of the data center. On the basis of improving the management of the airflow tissue, the research method of the end precision air -conditioning operation is performed. Through the measurement and optimization Relatively balanced distribution, establish management strategies for air -conditioning energy -saving operations optimized based on air flow tissue. Finally, the airflow tissue and air -conditioning energy -saving management strategy is intelligent, and precise control of measurement, optimization, management and energy saving.

The application of AI intelligent control technology helps improve the energy efficiency of data centers. It is mainly used in the power supply system and refrigeration air conditioning system. Increasing the use of renewable energy is a key step in reducing carbon emissions. Intel has developed a solution that can integrate into the existing energy network infrastructure to create a grid with a higher level of intelligent energy consumption and energy sources that can adapt to change. Intel joined some of the world's largest public business operators to form the Edge for Smart Secondary Substrations Alliance to achieve the modernization of power grid -changing power stations and better support renewable energy. ENEDIS, the largest power grid operator in France, recently joined this alliance and adopted solutions to provide real -time control of the entire network to upgrade more than 800,000 secondary substations. Beijing Jinfeng Huineng Technology Co., Ltd., which provides intelligent operations for wind power power plants, uses Intel AI solutions and CPU integrated AI to accelerate, and uses Intel® DLBOOST and Analytics Zoo to increase the accuracy of wind energy prediction from the original 59%to 79.41 %. Apply AI intelligent control technology to the automatic control logic of the refrigeration and air -conditioning system. By predicting outdoor weather parameters, analyzing load changes, calculating the characteristics of cold source equipment, the control logic of the cold source system, maximizing the outdoor natural cold source to achieve High -efficiency and energy saving of the refrigeration system.

High -voltage DC (HVDC) is a technology that has attracted much attention. Compared with the traditional UPS, a DC/AC inverter link is reduced, which greatly improves the efficiency of electrical energy utilization. Compared with the traditional UPS system scheme, its energy -saving effect can be improved at the highest calculation. 8%. At the same time, due to the small number of DC transmission wires, there is no contribution to no merit and no meritorious losses, only the heating loss of the resistor, which improves the energy saving effect in the power distribution transmission. DC transmission only requires two wires, which can save a lot of line investment, so the cable cost is much saved.

05

Improve energy efficiency from the organizational level

Factors affecting the energy consumption of data center involve multiple aspects, including server systems, air -conditioning refrigeration systems, power supply and distribution systems, computer room decoration, lighting, power distribution equipment, power supply and distribution cables, etc., and even the level of operation and maintenance will affect the data center Energy consumption. By monitoring and analysis of the operation load and energy efficiency of the data center, the energy consumption problems existing in the operation of each subsystem are found, and the operation adjustment and technical improvement of each subsystem will be targeted to reduce the data center to reduce the data center Electric energy consumption, improve the overall energy efficiency. Traditional data centers adopt an energy consumption analysis method combined with artificial or manual or dynamic loop system. It is difficult to monitor comprehensive monitoring, accurate prediction, analysis and control in the face of massive amounts of operation data and reports. Therefore, the data center energy consumption analysis tool must be used to control the energy consumption of the data center equipment, system -level, and project -level energy consumption.

For example, the data center energy consumption analysis tool developed by Intel mainly covers four functional segments: monitoring, analysis, prediction and control.

● Monitoring function: Automatic discovery function with the network, supporting a variety of equipment and multiple protocols, supporting various brands and equipment, can track the trend of energy consumption of equipment;

● Analysis function: Timely discover the hotspots in the machine room, have the server energy consumption analysis function, identify the uneven problem of air conditioning refrigeration and airflow tissue, and timely discover the "zombie server", have capacity analysis functions, realize intelligent capacity management;

● Forecast function: Vendent growth prediction, temperature health prediction, energy consumption and temperature prediction of associated applications;

● Control function: server energy consumption control strategy, based on hot refrigeration analysis, increase server cabinet capacity, and migrate based on temperature/power consumption application.

Based on the data center energy consumption analysis tool monitoring, analysis and prediction of data information such as the operation status and energy consumption of various equipment, systems, and energy consumption situations, problems are discovered in time, and accurate efficient energy -saving control strategies are formulated, relying on the control function of energy consumption analysis tools The implementation of the control strategy to achieve comprehensive monitoring, accurate prediction analysis and control full -chain energy consumption management strategies.

In the data center industry widely used CQC8302-2018 "Technical Specifications for the Operation and Maintenance Evaluation of Data Center Infrastructure", the following requirements are as follows:

1) Formulate the energy efficiency management system, clarify the collection requirements, collection methods and frequencies of energy consumption data, and systematically analyze the collection data. The analysis content includes but is not limited to:

Statistical analysis data center overall power consumption changes;

a. Statistical analysis data center daily power consumption and average power consumption;

b. Statistical analysis data center energy consumption composition and proportion;

c. Statistical analysis data center monthly energy efficiency indicators change;

d. Statistical analysis UPS system's own energy consumption changes;

e. Statistical analysis of the changes in energy consumption of the heating system of the air conditioner;

2) Statistical analysis data center water consumption changes (the selection of air -cooled system is not applicable) to understand the characteristics of the operating characteristics of IT devices:

a. Whether to analyze and understand whether the IT device running at the granular IT device carried by the data center infrastructure is analyzed and understood;

b. Whether to communicate with the relevant departments of customers or users, make predictions for the deployment of high -density IT loads, and formulate relevant response plans.

3) The ability to manage airflow tissue, including but not limited to: a. The possible leakage outlets of facilities buildings should be blocked to maintain the positive pressure of the facilities;

b. Practice the flow of airflow in the facilities, the possible leakage outlets of the blocking, install all the idle U -bit in the cabinet, close the unnecessary air outlet, and ensure the best use efficiency of cold air.

4) There is a management system and cyclical correction strategy for running threshold settings.

5) Based on the comprehensive consideration of safety and operating efficiency, establish a guidelines for the operation threshold setting, set up monitoring alarm thresholds, air conditioning return temperature, etc.

6) Regular energy consumption analysis meetings to continuously improve and optimize energy consumption management strategies.

If the above energy efficiency management requirements are operated by manual manual, it will be greatly consumed, and it is difficult to continue to meet the requirements. Intel's data center energy consumption analysis tool can help data center operation and maintenance personnel continue to meet the CQC8302-2018 "Data Center Basic Basics Basic Basis Requirements for the technical specifications of facility operation and maintenance evaluation.

The energy efficiency management of the data center project needs to be combined with the organization's energy measurement system and energy management system to integrate into the data center operation and maintenance system. On the basis of safe operation, we continuously seek energy conservation potential, and adopt various innovative technologies to improve the overall energy efficiency level.

06

Conclusion and Outlook

Although the energy consumption of data center is relatively large, the energy consumption level of the entire data center has not increased much compared with the computing power provided. Among them, it is mainly due to the improvement of server energy efficiency levels in recent years. A paper published in the journal of Science on February 28, 2020, the title is "Global Data Center Energy Use still slows down under demand rapid growth". The study of the paper calculated that from 2010 to 2018, the demand for global data centers increased by 550%, and the energy center energy use increased by only 6%. Therefore, data centers that have an important role in the digital economy still need to develop vigorously.

While vigorously building data centers, focus on the research and application of energy -saving and low -carbon green technology. At the same time, energy -saving and low -carbon technology needs to be organically integrated with management to jointly drive the efficient energy saving of the data center. "Three -point construction, seven -point management" is already the consensus of the data center industry. Through the application of new energy -saving technologies such as liquid cooling technology, high -efficiency cooling technology, and high -efficiency chip server, in conjunction with refined operation and maintenance management, the green center of the data center is finally realized. Energy saving.

The application of intelligent technology promotes the high -quality development of data centers. AI intelligent control technology can effectively improve server computing efficiency, power supply and distribution efficiency, and refrigeration energy efficiency, and promote the improvement of data center energy efficiency. With the help of intelligent energy consumption analysis tools, through the implementation of module functions such as monitoring, analysis, prediction, and control, the energy consumption monitoring and control of various equipment, subsystems and project -level energy consumption is achieved to help the data center green and efficient development.

Therefore, the development of green energy -saving and low -carbon development of data centers cannot simply pursue extremely low electrical energy use efficiency (PUE), and requires comprehensive considerations such as computing power, energy consumption of IT equipment, HVAC energy consumption, power supply and power consumption, and other aspects. Essence

Data center green and high -quality development has a social leadership and exemplary role. Not only will the high -energy consumption industry transform the development method quickly, accelerate the digital transformation process of thousands of industries, and achieve rapid energy saving and carbon reduction. Essence Through technological migration, these ideas can help other high -energy consumption industries, including agriculture, logistics, mining, and manufacturing to achieve reducing carbon emissions. While these industries realize the digital transformation, accelerate the pace of carbon neutralization and transformation, thereby therefore Realize the high -quality development of data centers and give full play to greater social value.

Author: Wutong, Director of the Advanced Surveying Engineering Center of China Institute of Metrology Science

- END -