There are granaries for grain, and reservoirs for water.

In the era of digital economy, people's daily work, study, social interaction, consumption, transaction and other behaviors will generate massive data, and the storage and calculation of these data cannot be separated from the "granary" and "reservoir".

As a "data granary", the data center not only supports the digital economy accounting for nearly 40% of the country's GDP, but also brings Chinese technology and Chinese products to the big stage of overseas markets.

  What does the domestic "data granary" look like?

How are they built and operated?

How is it performing in overseas markets?

  "Air conditioner" must use green and low-carbon

  On a winter night, in Huailai County, Hebei Province, located at 40 degrees north latitude, the cold wind howled.

Wang Dongsen was fully armed, came to the top floor with deicing tools, and walked quickly to the cooling tower.

  "It's too cold at night, and the hot air radiates from here, and it freezes when it's cold. If it's not dealt with in time, it will affect the heat dissipation and may cause malfunctions." Wang Dongsen told reporters.

  Wang Dongsen is an operation and maintenance engineer in the data center of the headquarters base of Chindata Group. He and more than 100 operation and maintenance colleagues guard the servers in tens of thousands of cabinets in the data center of the headquarters base around the clock.

Thanks to their careful care, the high-density servers are running in an orderly manner, and users who are thousands of miles away watch videos and watch live broadcasts, enjoying a smooth experience.

  It was cold and windy outside, and the servers and fans were humming in the computer room.

  "Cooling is the most important thing for the safe and stable operation of the data center." said Han Baoliang, Director of Operation and Maintenance of Qinhuai Data Center.

"It's like playing games on a mobile phone. After a long time, it will be hot and freeze. When the server is running at high frequency, it will also generate huge heat. If the heat dissipation is not good, it may be 'downtime'. We must find a way to cool down the server."

  If the home is hot, you can turn on the air conditioner. When the data center is hot, you can’t do without the air conditioner. However, the data center is a big power consumer. To cool down the server, you must use green and low-carbon air conditioners to reduce power consumption.

  Han Baoliang told reporters that site selection is the key. In addition to safety factors such as being far away from earthquake zones, climate factors must also be considered. Choosing an area with a cool climate can efficiently utilize natural cooling sources and reduce operating costs.

In Huailai, a large number of leading enterprise data centers have settled down, which is inseparable from the cold climate.

  On the other hand, data centers can achieve better cooling effects by comprehensively using a variety of advanced cooling technologies in different computing power scenarios and business models.

  There is an energy consumption evaluation index PUE in the industry, that is, power usage efficiency, which is the ratio of all the energy consumed by the data center to the energy consumed by the IT load.

The closer the PUE value is to 1, the less energy is consumed by non-IT equipment, that is, the higher the energy efficiency level.

  "Our headquarters base data center uses technologies such as indirect evaporative cooling, water cooling, and water-free cooling. The cooling tower is a part of the water cooling technology." Han Baoliang said, "The annual average PUE of this park reaches 1.17, and the energy efficiency level is much higher than the industry average. Level."

  Another example is Tencent.

After years of accumulation, Tencent Cloud Data Center has completed four generations of technology iterations. The latest generation of T-Block data center independently developed and owned by Tencent Cloud with complete intellectual property rights has a PUE of no higher than 1.3, which can save electricity for a park with 300,000 servers every year. About 250 million kWh.

  The server consumes the most power, but the PUE value cannot be reflected.

With strong R&D capabilities, Tencent reduces the energy consumption of its self-developed servers, and at the same time uses big data algorithms to optimize server scheduling and operation methods, improve power consumption efficiency, and make data centers more green and energy-efficient.

  The data center industry is developing rapidly, and the green and low-carbon wind is blowing.

Third-party manufacturers such as Chindata and 21Vianet, Internet companies such as Tencent, Alibaba, and Kuaishou, and operators such as China Telecom, China Mobile, and China Unicom, are all speeding up the pace of greening data centers by relying on technological means.

  Data show that in 2021, the national data center power consumption will reach 216.6 billion kwh, accounting for about 2.6% of the country's total power consumption, supporting the digital economy that accounts for nearly 40% of the country's GDP.

  "my country's early data center construction was relatively extensive, and the utilization rate and energy efficiency of data centers were low. In recent years, under the guidance of national policies and the promotion of platforms such as the Open Data Center Committee (ODCC), my country's data centers have gradually entered a stage of high-quality development." Li Jie, deputy director of the Cloud Computing and Big Data Research Institute of the China Academy of Information and Communications Technology, said, "By adopting new energy, liquid cooling, high-voltage direct current and other technologies, deploying an intelligent management platform and improving the refined operation and maintenance management system, my country's data center The level of energy efficiency has been greatly improved, and the operation management and service capabilities have been continuously improved to better meet business needs."

  Not afraid of sudden power failure

  In daily life, insufficient power supply or line failure during the peak period of electricity consumption will cause temporary power outages.

The data center consumes a lot of power and often runs 24 hours a day, so it is inevitable to encounter various emergencies.

  "What should I do if there is a sudden power failure?"

  Facing the reporter's question, Han Baoliang was very confident: "We have double insurance, the battery system and the diesel generator system ensure the stability of the circuit to the greatest extent."

  When I came to the power distribution room of the data center of the headquarters base, the reporter saw dense blue batteries neatly stacked on rows of iron frames. These batteries were first connected in series and then in parallel, and connected to the uninterruptible power system. , the battery pack here can be switched immediately and continue to supply power. IT equipment will not have any perception of this instantaneous conversion, and it is difficult for end users to detect it." Han Baoliang said.

"But the battery system can only supply power for 15 minutes, and the automatic control system will start the diesel generator within 10 seconds to restore continuous power supply, and the whole process does not require human intervention."

  "It takes at least 30,000 steps to make a round of the computer room inspection in the park. In addition to routine manual inspections and daily drills, we have also independently developed the Kunpeng IDC operation platform to monitor the operating status of the system and equipment in real time. The automatic response mechanism can also realize Quickly discover, analyze, and handle faults." Han Baoliang said, "As of July last year, the total monitoring points of the Kunpeng platform were nearly 10 million, and the daily processing data volume exceeded 5 billion."

  In the digital economy era, data center operation and maintenance has also moved from manual to automated.

Using digital twin technology, Tencent projects hundreds of data centers 1:1 into the digital full-reality scene to realize centralized management of ultra-large-scale data center clusters.

  "If any device is abnormal, it can send an alarm to the cloud centralized monitoring center within a few seconds to improve processing efficiency." Yue Shang, director of Tencent Cloud Data Center Zhiwei Platform, introduced.

While ensuring reliability, data center operation and maintenance costs are saved by more than 80% compared with traditional models, and labor costs are saved by more than 50%.

  Capacity management, equipment health management, energy consumption management... new intelligent operation and maintenance tools have become the intimate "little padded jacket" of operation and maintenance personnel.

  "my country's data center construction and operation companies have made a lot of technological innovations in data center facilities and IT equipment. Every year, the Open Data Center Summit releases about 50 technical specifications and white papers, which play an important role in the application and promotion of these innovative technologies. The application of these innovative technologies has strongly promoted the development of my country's data centers in the direction of high technology, high computing power, high energy efficiency, and high security." Li Jie said.

  Li Jie believes that overall, in terms of management and application, there is still a certain gap between China's data centers and the international leading level.

In recent years, data centers have continuously increased their attention to operation and maintenance management, and intelligent management has accelerated the development. However, operation and maintenance management technologies, systems, systems, and personnel still need to be systematized, standardized, refined, and intelligently constructed and optimized.

  Chinese products and technologies gradually go to the world

  Against the backdrop of the rapid development of the global digital economy, the competitive arena for data center companies continues to expand.

There are both "soft" services and "hard" equipment. Chinese data center companies are showing their talents in overseas markets.

  "Chinese enterprises have a huge demand for 'going overseas', and the construction and improvement of digital capabilities are essential." Guo Songtao, overseas resource planning manager of Tencent Cloud, introduced that according to traditional methods, enterprises need to build their own data centers and deploy underlying computing, storage, and network resources. To ensure the normal operation of the system through daily operation and maintenance, the capital cost is high and the time is long.

  This is the advantage of data center cloud service providers.

"Tencent has many years of experience in 'going overseas' and can provide faster and more efficient localized services overseas." Guo Songtao said, "For corporate customers going overseas, they only need to deploy various software on Tencent Cloud Services."

  Taking Glory as an example, in the process of serving global consumers, cloud deployment must achieve high stability, high reliability and low cost.

"To ensure that consumers can get a consistent experience no matter where they are, this requires the strong support of global infrastructure." Wang Weipeng, Senior Account Manager of Tencent Cloud Consumer Electronics Industry, introduced.

  Relying on Tencent Cloud's global data centers, Honor completed the migration of the first batch of overseas business systems in just 15 days, which greatly shortened the time to go online.

  Operating more than 100 data centers in 26 geographic regions around the world and deploying more than 2,800 CDN acceleration nodes—Tencent Cloud helps overseas companies ride the wind and waves with its high-quality services in elastic computing, storage, big data, AI, security, etc.

  As a third-party manufacturer, Qinhuai Data has drawn different "going to sea" trajectories.

  In October last year, in Johor, Malaysia, a hyperscale data center went up.

It took only 11 months from wasteland to delivery, much lower than the construction time of local data center projects.

This is the first phase of the Malaysia Johor Hyperscale Data Center (MY06 for short). The project is divided into 3 phases. After completion, it will be the largest hyperscale data center in Malaysia.

  Fast delivery, thanks to the new construction model.

All prefabricated modules of the first phase of MY06 are designed and developed by Chindata, prefabricated in China, integrated and assembled, and then delivered overseas as a whole.

At the construction site, building a data center is like "building blocks", and the speed is naturally greatly improved.

  According to Li Mingjiang, the person in charge of the project and vice president of Qinhuai Data Group, this is also China's first overseas practice of ultra-large-scale prefabricated data center technology and products. The project has obtained 15 technical patents and 1 software copyright.

It is worth mentioning that the design PUE of the first phase of MY06 project is lower than 1.2, which is a new breakthrough in energy efficiency in tropical areas, which is not easy in Malaysia, where the climate is hot and humid.

  "Our country has a wide geographical range, rich and diverse climate and environment, and has accumulated rich experience in data center construction and operation, forming a large number of products and technologies with Chinese characteristics and regional characteristics." Li Jie said, "In the future, we need to further deepen original technology innovation. , optimize product quality, aim at key overseas markets such as the 'Belt and Road' and Southeast Asia, gradually promote power supply, refrigeration, IT, network and other technologies, continue to accumulate and explore, and then gradually promote Chinese products and technologies to the world."

  further reading

  What is the relationship between big data, cloud computing, and data centers

  Big data is an application scenario under the background of the mobile Internet and the Internet of Things. The huge amount of data generated by various applications needs to be processed and analyzed to mine valuable information.

  Cloud computing provides solutions for big data processing and value mining.

Big data cannot be processed by a single computer, and needs to rely on distributed processing of cloud computing, distributed database and cloud storage, virtualization technology, etc.

  The data center is the infrastructure of cloud computing.

Data centers are distributed in different core cities and radiate to surrounding cities to provide basic support for network communications.

Cloud computing is a utilization of idle resources from basic services to value-added services provided on the basis of data centers.

No matter how cloud computing changes, it must rely on the data center to achieve landing.

  It can be said that the data center is the root of cloud computing, and cloud computing is the "leaf" of the data center. Cloud computing promotes the development of data centers through "photosynthesis", and the growth of data centers provides a solid foundation for the development of cloud computing.

Big data, cloud computing, and data centers are interdependent and mutually reinforcing.

  (Compiled by our reporter Kang Pu)