Data resourceization is a prerequisite for data socialization, and computing is the key bearing of data resourceization. Computation + will become the driving force behind the evolution of data engines.
Data is encircling and transforming the world. According to market research firm App Annie, the global data holdings will reach 35ZB in 2020. In today's increasingly large data stocks and increasing computing pressure, will computing become a bottleneck for mankind to move toward a smart society? At the 2016 Inspur Technology and Application Summit, Wang Endong, Executive Director of Inspur Group and Academician of the Chinese Academy of Engineering, clearly stated in his keynote speech: “Data resourceization is a prerequisite for data socialization, and computing is a key carrier of data resourceization, computing + Will become the driving force behind the evolution of the data engine."
Big data explosion, computing power will become the power source of data socializationIn his keynote speech, Wang Endong portrayed the industry background of the rapid expansion of data resources and the socialization of data. Just as smart terminals are rapidly urging the mobile Internet industry, massive sensors have become a huge data torrent of the Internet. According to BI Intelligence's research report, by 2020, there will be up to 34 billion devices connected to the Internet, and IoT devices will reach 24 billion. According to market research firm App Annie, the number of smartphones and tablets in the world will reach 6.2 billion units by 2020, a huge increase of 139% compared with the 2.6 billion units in 2016.
These new mass devices are different from the past. Most of them will be connected to the Internet, and more amazing daily data will be born every day. When the amount of data and equipment generated by each device in the multiplication increases sharply, then there will be a multiplication. An exponential surge in terror may be formed. Wang Endong calculated a data account on the spot. Now an intelligent terminal has an average of 16 sensors, which can generate 1GB of data per day. If the calculation is based on 2.6 billion smart terminals worldwide, the amount of data added in one day will be as high as 2.6EB. At present, nearly 3 million surveillance cameras are installed in the Chinese public security organs, and the data is generally stored for 30 days, which will generate huge data of 3EB. If 340 billion devices are connected to the Internet in 2020, the amount of data generated by human society will be as high as 35ZB (1ZB data, which requires 2 30-bit ITB hard drives to be installed).
Some people regard big data as the most important means of production in the future, and their future strategic value even exceeds that of oil. But just like putting a lot of delicacies in front of a gourmet, the limited digestion of the stomach means being powerless. Similarly, excessive data will also bring about the problem of "indigestion". At present, the world's fastest Shenwei·Taihu light has a peak computing power of 125 PFlops, and the E-level calculation needs to be improved by nearly 100 times, so when faced with ZB At the level of data, the increase in computing power becomes imminent.
In Wang Endong's view, in this process, data is collected through the smart device at the front end, and the data is pre-processed to form valuable information, which is transmitted to the back-end data center through the network for storage, modeling and statistical analysis to form knowledge. Further, we provide services for our work and life decisions through in-depth analysis and mining such as classification, induction, deduction and prediction. Every aspect of this is done through calculations. Therefore, calculation is the key to data socialization.
there is always a solution to a problem. Some people may say that since the calculation is so important, then we can improve the computing power, but in fact, the difference between typical applications and typical data, the requirements for computing are also different, so the path to calculate upgrades is also different.
In an interview, Wang Endong said that from the single numerical calculations in the past, with the continuous maturity of various calculations and the continuous enhancement of computing power, the calculations have evolved to today, and finally formed the key business calculations, intelligent calculations and scientific calculations. In the typical form, these three forms coexist for a long period of time, and their application fields and corresponding data types also have their own characteristics, so the improvement of computing power needs to be considered from the origin of requirements and applications.
Key business calculations, in fact, everyone can feel the convenience brought by them in life. The online banking payment we made during online shopping, WeChat payment after 711 is bought early, the travel to the field is to use 12306 to book high-speed rail. Tickets, these are actually key business calculations, such calculations require ultra-high stability and security, so the future development focus is to enhance high performance and high reliability.
Smart computing is mainly used in the Internet of Everything. Tens of billions of sensors are used as data acquisition sources. Together with the Internet, we have woven a grand map of intelligent transportation, smart medical, smart city, social network and artificial intelligence. Wang Endong pointed out that smart computing, represented by cloud computing, big data and deep learning, will increase its market share in the next few years, and it may reach a share of around 1/2 in 2015 and reach 2/ in 2020. 3, even in 425 will reach 4/5, becoming the most important part of future computing.
Scientific computing has become an important tool for human exploration of the unknown. In the past, it was often said that the secret can not be leaked, but now human scientists use supercomputer to make weather forecast and meteorological disasters are very mature, but it is the limitation of computing power that hinders the further improvement of prediction accuracy. For example, to achieve 200 km grid resolution accuracy simulation, it requires billions of calculations, but if you want to achieve 25 km grid resolution simulation, it needs 10 billion times.
If you look at the unknown universe, the pressure of scientific computing will increase further. At present, the world's largest radio telescope array SKA, used to detect galaxy evolution, gravitational waves, etc., will generate 1.5TB of data per second, 129 PB of data per day. The huge increase in data requires the matching of computing power. Support, to a certain extent, human computing power is the boundary of the cognitive world.
From some typical applications mentioned above, we can see that the key business calculations usually correspond to structural data, and most of the data growth is linear growth, while intelligent computing and scientific computing provide an index for the improvement of computing power. The demanding requirements of the class upgrade. Today, when Moore's Law ends, the performance of a single compute node is limited, and the power challenge is increasing, how does the evolution evolve? The solution given by Inspur is calculation + strategy.
Calculate +, use fusion architecture to achieve technology and business restructuringHow to realize the re-evolution of computing under the condition of limited restrictions? In an exclusive interview, Wang Endong said that in order to further tap the computing potential under the existing technology and engineering conditions, the fusion architecture is an inevitable technology choice. In the past, it relied too much on the performance upgrade brought by the chip. Now it needs to go from the architectural innovation level. Re-cognition. In the understanding of Inspur, the converged architecture is the application-oriented hardware refactoring + software definition.
The design idea of ​​the converged architecture is to realize the physical pooling and dynamic reconfiguration of resources through hardware decoupling, and realize the service-aware on-demand resource combination and configuration through software definition, to meet the elastic expansion and expansion of the computing system and the super-large-scale continuous expansion. The convergence architecture will not only change the technical architecture of the data center, it will even lead to the integration and reconstruction of the industry. Now, the integration of the ICT industry is actually happening.
In reality, following hardware refactoring + software definition, computing +SDN/SDS has already implemented preliminary computing, storage, and network convergence in deployment. Users will not distinguish between server vendors or network vendors because they need The solution is a package, the original dedicated equipment will be less and less, and after the hardware is guaranteed to be versatile, most of the functions of the dedicated device can be realized by adding custom function modules and software.
Inspur has cooperated with Cisco's products and technologies in NFV, SDDC, IoT and other fields, and deployed in typical applications such as information infrastructure, cloud centers, and smart cities. In my opinion, computing, storage, and networking are becoming more dependent and indivisible in the future.
According to Academician Wang Endong, Inspur has fully prepared for the implementation of the integration architecture: Inspur for key business computing, already has K1 host, AS18000 storage system and K-DB relational database, and will release a new generation this year. The host system; on the converged architecture modular server, the Inspur SR cabinet has been changed to 4.5, from the non-IT module centralized development to the hardware IO pooling and storage pooling, creating a new record of 10,000 servers on the line one day; To meet the strong demand for deep learning applications, Inspur uses a co-processing optimization node that can provide 160 GPUs or MIC cards in a single cabinet, which is four times more dense than traditional solutions. In addition, Yunhai OS is co-ordinated and managed at the software definition level, so Inspur is confident that the computing can be re-evolved through the fusion architecture.
The fusion architecture is inseparable from the strong "friend circle" supportIntegration is a major road to future computing and even technological advancement. However, integration has also caused a serious impact on the original business structure and distribution of interests. So for the wave of the concept of fusion, it is extremely important to make your own fusion structure friend circle more powerful. Inspur has already cooperated with Intel, Cisco, VMware and other international giants on the converged architecture, but in the future, it needs to introduce more and more partners, so that typical users can really put down their concerns and embrace the converged architecture.
The data socialization proposed by Inspur this time uses calculations to promote data resourceization. In fact, it is the same as the social data physics proposed by Alex Pentland, the global big data. Professor Pentland's social physics mainly emphasizes the flow of ideas and information beyond material and energy. The core tool for human society is the big data, and the calculation + is exactly The key prerequisite for data resourceization.
In the past, computing power was a very important productivity, but now with the exponential demand of smart computing and scientific computing, computing power has become an important material basis for innovation. Computational power is taking on the main engine role of human technology evolution. Imagination and curiosity allow the human cognition world and the boundaries of the universe to evolve, and the computational power evolves to determine the actual boundaries of this extension.
Plastic Boxed Access Control Power Supply
- Plastic case power supply
- Battery over-charge, over-discharge protection
- Additional current battery charge
- Pro-wired battery connection
- Product description: Applicable to all types of Alarm monitoring systems, Access control systems, Building intercom systems
Input voltage: 100-240V AC @ 50/60Hz
Output voltage: 13.8V DC
Output current: 2A .3A. 5A
Charge current: 1.0A Max.
Ripple&Efficiency: <100mV / >80%
Recommended battery: 12V7Ah , 12V 17Ah
Access Control Power Supply, Boxed Power Supply UPS,Power Supply for access control system
Guangdong Steady Technology Co.LTD , https://www.steadysmps.com