豆奶抖音视频app

Welcome to the official website of Xiangjiang Technology Co., Ltd.
Advisory Hotline: 4008-6044-55 | OA | E-mail
The visible future of 2019: five new trends and five technology solutions for the data center market
Author: Anonymous Source: China IDC Circle Views: 1909 Date: 2019/1/30 8:48:37

5G, Internet of Things, Made in China 2025, and artificial intelligence technologies are becoming important forces driving innovation and achieving digital transformation. A new digital economy era is coming. At the same time, the rapid development of smart grids and microgrids, as well as the continuous reduction of alternative energy costs and energy storage costs, have promoted more interaction and integration of data center physical infrastructure with the upstream grid and downstream IT. As a global leader in the field of data center infrastructure construction and services, Schneider Electric is keen to grasp market changes and customer needs based on deep industry insights and practical experience to continuously innovate physical infrastructure, software and digital solutions, and whole life Periodic services provide guarantees for the construction and operation and maintenance of high-availability digital cornerstones. Looking forward to 2019, Schneider Electric believes that the data center will usher in the following five new trends and five technical solutions:

One of the new trends: edge computing and cloud computing will develop synergistically

In the past year, we have seen a lot of hype about edge computing, such as: "The cloud center is dead, edge computing is superior?", "Edge computing will change the future of the Internet and the Internet of Things", "Edge computing will lead the next IT Change "and so on. But we don't think cloud computing will die out, and edge computing will not replace cloud computing. Rather, edge computing and cloud computing will develop together. With the rapid development of the Internet of Things, AI, and 5G, people have higher and higher requirements for latency and broadband. Edge computing is specifically designed for this "speed and bandwidth need."

Schneider Electric believes that the future IT architecture will be a hybrid IT architecture that will consist of three types of data centers, each of which is a central cloud data center located in a remote area for ultra-large-scale computing and storage. Area edge data centers near users for large-scale computing and storage, and edge data centers located locally near data generation and use locations. Edge data centers will solve the fast response problem of computing locally, and cloud computing will provide powerful back-end computing support and data storage capabilities for edge computing. At the same time, we also believe that edge computing data centers will first develop rapidly in the two major areas of the retail industry and network operators to meet people's needs for low latency and high bandwidth.

New trend two: the era of liquid cooling is coming

Air-cooled IT occupies a very important position in data centers. Today, air-cooled IT is still the main data center. However, the application of artificial intelligence places huge demands on the processing capabilities of data centers. Artificial intelligence has begun to leapfrog development, and has developed from laboratory research to real business and consumer applications. These applications are so computationally intensive that many IT hardware architects started using GPUs as core processors or auxiliary processors. Many GPU-based servers, or what we call TDP-Total Design Power, typically generate around 300 watts of heat, compared to 130 to 150 watts for traditional servers. GPU-based servers generate twice as much heat as traditional servers, which is one of the main drivers of the rise of liquid cooling.

According to the survey, we found that the power density of the current cabinet in the data center is about 7 kilowatts, but the power density of each cabinet may be as high as tens of kilowatts in the future. We think that if the power density of the cabinet is less than 20 kilowatts, the traditional cooling system is very cost-effective. However, when the power density of the cabinet exceeds 20 kWh, the traditional air cooling technology will face challenges such as air volume, power consumption, and noise. At this time, liquid cooling technology is required.

Especially in the first-tier cities (such as Beijing and Shanghai), government management departments have increasingly strict requirements for data center energy consumption and emissions per unit of output value. On September 26, 2018, the `` Restrictions and Prohibited Catalogs of New Industries in Beijing 2018 Edition '' requires that new and expanded data centers for Internet data services / information processing and storage support services be banned in central urban areas and Beijing sub-centers. New cloud computing data centers with PUE below 1.4 are allowed. On October 29, 2018, the Shanghai Municipal Commission of Economics and Information Technology issued the "Shanghai Three-Year Action Plan for Promoting the Construction of New-Generation Information Infrastructure to Boost Urban Energy Levels and Core Competitiveness (2018-2020)", and proposed that the new data center PUE must be low At 1.3, the PUE of the existing data center is not higher than 1.4, and at least 50% of the new data is AI-oriented. At this time, the traditional air cooling method is difficult to achieve these PUE requirements, which will also become an important driving factor to promote the rise of liquid cooling.

New trend three: lithium batteries will have more applications in data centers

We started using lithium batteries in digital DV more than two decades ago, but why isn't it popular in the data center field? The main reason is that lithium batteries cannot provide UPS suppliers with a balance of price, energy density, safety and reliability. However, in recent years, the rapid development of electric vehicles and new energy has promoted the advancement and price reduction of lithium battery technology. Today, the cost of lithium batteries has decreased by 70% compared with 5 years ago. At the same time, the chemical cost and technology of lithium batteries have improved It also provides realistic solutions for suppliers.

Compared with traditional lead-acid batteries, lithium batteries have excellent characteristics, including higher energy density (70-260 kWh / kg vs. 15-50 kWh / kg), longer life cycle (10-15 years vs. 4-6 years), fast charging (1 / 2–1 hour vs. 6-12 hours), occupies 50-80% of land saving, 60-80% of weight reduction, charging and discharging times> 1000 cycles vs. <400 Cycle, etc., will gradually replace the traditional lead-acid batteries as the backup battery of choice for data center UPS. According to a global data center survey conducted by Uptime in 2017, 10% of data centers have adopted lithium batteries as a backup energy source.

As the cost of microgrids, smart grids, new energy, and energy storage is reduced, and many countries and regions have higher requirements for the sustainable development of data centers, more and more data center owners are beginning to consider Through lithium battery energy storage to reduce costs and achieve the goal of sustainable development of data centers. As the world's largest lithium battery producer, China's excess lithium battery capacity also needs to explore new application growth points, such as wind and solar energy storage, grid frequency and peak modulation, and communication power. Therefore, we believe that lithium batteries will be used more and more in data centers.

Fourth new trend: The value of data center management as a service (DMaaS) in data center planning, construction and operation is becoming more prominent

Data center infrastructure management (DCIM) system was originally designed to collect information from the physical infrastructure equipment of a single data center and was deployed as a local software solution. The new cloud-based management solution will be located in the cloud, which can help users collect massive data from a wider range of IoT-based devices, and the software management platform can implement a single, hundreds, or thousands Physical infrastructure management for large or small data centers.

These new systems are often referred to as Data Center Management as a Service (DMaaS). They can help users make predictive decisions through big data analysis, thereby reducing unexpected events or downtime. At the same time, It is also much faster than traditional DCIM solutions. As a cloud-based management software, the software can use the "data resource pool" to collect and store data and analyze trends, which can help data center users to plan their business from a strategic level.

The cloud-based management software also simplifies the requirements for new equipment deployments, as well as tasks for upgrading existing installations, and software updates can be performed at multiple different locations. Because it is very challenging to manage and upgrade the sites one by one, especially for edge data centers, using local management software.

Fifth new trend: the construction of very large-scale data centers requires increasingly high deployment cycles

We believe that the demand for cloud computing in the data center market will neither weaken nor slow down. It will accelerate development in 2019, which means that network giants or managed service providers will add more computing capabilities by building very large-scale data centers. However, the data center market will drive them to build these facilities faster and faster, that is, 10 to 100 MW data center projects need to complete from design, construction to operation in less than a year.

In order to solve the various challenges faced by users in a constantly changing market environment, data center infrastructure solution providers with Schneider Electric as the leader will provide users with the following five technical solutions through continuous technological innovation:

One of the technical solutions: edge computing solutions

With the development of edge computing, Schneider Electric believes that the past understanding of the availability of a single data center is insufficient. We need to change our mindset and look at the issue of availability from the perspective of user experience. As long as the user's application or network is interrupted For example, applications that cannot connect to the APP should be called faults. Therefore, we need an edge data center with high availability that should include air-conditioning, UPS, monitoring, IT equipment, network, storage and other equipment.

At the same time, we believe that to achieve large-scale deployment of edge data centers, we need to solve three problems. First, we need to establish a system that includes users, system integrators, physical infrastructure providers, IT equipment manufacturers, and managed service providers. Collaborative ecosystem; second, we need to deploy cloud-based management tools to achieve intelligent management of hundreds or even thousands of edge data centers; third, we need to ask technology for big data analysis and artificial intelligence Human resources to solve challenges such as insufficient human resources to achieve unattended operation and maintenance management.

Technical solution two: liquid cooling solution

At present, there are mainly two liquid cooling solutions, one is chip-level liquid cooling, and the other is submerged liquid cooling. Chip-level liquid cooling is sometimes referred to as cold plate liquid cooling. These two types of liquid cooling have their own advantages and disadvantages. The cold plate type can refurbish most of the traditional servers. The disadvantage is that other components still require traditional room air conditioning for cooling. The advantage of immersion is that it eliminates the need for traditional server fans, but requires a redesign of the server motherboard.

In order to quantify the comparison between these two liquid-cooled and traditional chilled water systems, Schneider Electric analyzed the investment costs and operating costs from the perspective of data centers and buildings. It is found that the submerged liquid cooling is superior to the cold plate liquid cooling in terms of TCO. If viewed from the energy saving perspective of the refrigeration system, the cold plate liquid cooling can save more than 15% and the submerged liquid cooling can save more than 57%. From the perspective of energy consumption of the entire data center, the cold plate type can save 5% of annual energy consumption, and the immersion type can save 15%.

Coolant is very critical for the application of liquid cooling. We think that the discussion about coolant mainly focuses on the price of the coolant, compatibility with IT equipment, heat transfer characteristics, environmental impact, and the need to Think about issues from a life cycle perspective. If you want to deploy large-scale liquid cooling in the data center, you must overcome several issues including the need to redesign the server motherboard, and the other is the need to redesign the data center from a system perspective, especially the data In the central power supply and distribution system, how to design the power supply and distribution when the power density of the cabinet is as high as tens of kilowatt hours is a very challenging thought.

Technology Solution Three: Lithium Battery Technology Solution

The energy density of lithium batteries is more than three times that of lead-acid batteries. Without reasonable electrical protection design and BMS management, the consequences of out-of-control lithium batteries will be far greater than that of lead-acid batteries. However, we can solve this problem through the following four aspects: first, to ensure the high quality and high consistency of the battery cells through high-quality production process control; secondly, to achieve the charge and discharge of the battery cells through a complete BMS design Third, choose the appropriate lithium battery chemical materials, such as lithium iron phosphate, lithium manganate, nickel cobalt manganese ternary technology. Fourth, the lithium battery and UPS are tested for compatibility.

In the cloud computing mode, the power consumption of the server will be more dynamic. Ten years ago, when the server was fully loaded and idle, the power consumption was about 1: 2, and now the ratio between them can be as high as 1: 9. When designing the power supply and distribution system, the investment cost is huge. Therefore, many data center designers are discussing whether software and energy storage can be used to achieve peak shaving / deburring, so as to reduce the power supply and distribution system. design. At the same time, operating costs can also be reduced through energy storage and peak-to-valley price differentials.

The investment cost of diesel generators accounts for more than 10% of the data center physical infrastructure investment costs. At the same time, users also face challenges such as noise, pollution, and long deployment cycles. Investigation and research found that the average running time of diesel engines is 173 minutes, which means that we have more than 50% of diesel engines running time less than 3 hours. Schneider deduced that as the price of lithium batteries continues to decrease, by 2020, the 2.5-hour lithium battery backup system will have the same investment cost as Diesel Engine. Therefore, we believe that the lithium battery energy storage container will partially replace Diesel generators are also feasible.

The key to the energy storage of lithium batteries lies in the management of the battery system and the control of the software. Lithium batteries can implement each battery cell, module, rack and system through its own three-level BMS (battery management system). For online intelligent management, data center operation and maintenance personnel can understand the electrical status and health trend of the battery in real time through the BMS. Software management will be the key to the application of wind and solar energy storage, grid peak shaving, and UPS backup power in data centers.

Technology Solution 4: DMaaS Solution

In essence, data centers are very complex systems because they need to manage the interaction of power and cooling infrastructure with IT equipment while minimizing total operating costs. The need for data centers to continue running and minimize downtime continues to increase. When a problem occurs, we expect the management system to provide the necessary critical information to resolve the problem and analyze the root cause to prevent it from happening again. It can be seen that the industry has evolved into two data center management methods. One approach is to use "area-specific" systems for power, cooling, and white zone management. The second method is more "native", where SCADA-based systems are custom-programmed to perform management duties. Each method has its advantages and disadvantages, however, we believe that both methods will face limitations in the future and will require new approaches to address them to fully realize the potential of cloud-based tools and big data analytics.

As an early participant in the DMaaS market, Schneider Electric can use its data center years of practical experience to mine a large amount of data, including designing and building data centers, building management, power distribution, and power and cooling services. Schneider Electric's EcoStruxure architecture for data centers is an open, interoperable, IoT-based system architecture and platform, which can provide data center customers with more security, reliability, efficiency, sustainability, and connectivity. high value. And through the development of advanced Internet of Things, mobile, sensing, cloud, analysis, and network security technologies, it has realized comprehensive innovation in three levels of interconnected products, edge control, and applications, analysis, and services.

Interconnected products are the foundation of the intelligent operation of the Internet of Things. Schneider Electric's first belief is to take innovation as the core and produce great interconnected products. At the interconnected product level, data is collected from the various connected sensors in the data center and surrounding buildings, and from many vendors' equipment (UPS, PDU, power distribution and cooling systems, etc.). At the edge control level, whether at the building level, or in the IT system or power distribution system, EcoStruxure Edge Control enables customers to implement local control and monitoring, and autonomously monitor and take action on any alarm. At the application, analysis, and service level, data can be analyzed through remote digital platforms, and information can be extracted using artificial intelligence to provide actionable intelligence for preventive maintenance, energy efficiency, and other tasks.

Even if the data center owner uses a variety of equipment and software platforms, EcoStruxure can still read all health and status data. The data center owner can then share this data with the customer. Through the EcoStruxure architecture and platform, Schneider Electric will provide customers with high availability and efficient products and services, drive more value through predictive analysis and actionable intelligence, optimize business, and bring more returns.

Technology Solution No. 5: Hyperscale Data Center Solution

The use of prefabricated and modular skids will be the key guarantee for realizing the high requirements of the construction period for very large-scale data centers. The power supply and distribution skids prefabricate MW-level UPS, switch cabinets and management software in the factory in advance. These prefabricated modules are pre-tested and can be reliably and "plug and play" deployed after they arrive at the data center construction site. Because in some areas, the supply cycle of these power supply and distribution equipment can be as long as 12 months, and the use of prefabricated modular solutions can achieve timely delivery, eliminating any possible delays in the design and construction stages.

At the same time, it can also be assumed that the computing power of the data center will become more modular, enabling rapid on-demand deployment. Schneider Electric has developed a support system module deployment model that is deployed simultaneously with the cabinet implementation, and has been used by many hosted data centers around the world. In this solution, the infrastructure-independent frame structure can quickly push cabinets pre-installed with IT equipment into the aisle closed system, greatly reducing the time and complexity of deployment.

Online message

online service