Development plan for the integration of computing power network and media technology

Development plan for the integration of computing power network and media technology

Abstract: The computing power network is mainly based on the network means to realize the coordinated scheduling of resources (networks, computing, storage) to achieve the optimized and utilization of resources. At the same time, it provides users with a consistent business experience as the main development direction of the next generation of mobile Internet. The traditional media industry is facing the opportunities and challenges brought by the development of computing power network technology. In order to accelerate the integration and development of new computing power networks and media technology, strengthen the cultivation and incubation of new computing power networks, and discuss the computing power network technology and media technology technology Integration development.

Keywords: computing power network; media technology; integration and development; optimal resources; planning suggestions

Doi: 10.19483/J.CNKI.11-4653/N.2022.06.005

The record format of this article: Fu Wenjun, Wu Haiming, Zhang Hai, Xu Yan. Computing power network and media technology integration development plan [J]. China Media Technology, 2022 (06): 22-24.

About the author: Fu Wenjun (1988-), male, Guangxi Beihai, engineer, senior expert in Yunwang Technology, Zhejiang Mobile Information System Integration Co., Ltd., research direction: edge computing and computing power network; Wu Haiming (1987-), male, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang Tongxiang, Zhejiang, Tongxiang, Zhejiang, Engineer, Jiaxing Mobile Passenger Solutions Manager, research direction: DICT and 5G; Zhang Hai (1991-), male, Jiangsu Taizhou, engineer, engineer of Zhejiang Mobile Information System Integration Co., Ltd. ; Xu Yan (1981-), female, Hangzhou, Zhejiang, senior engineer, supporting supervisor of Hangzhou Mobile Information Operations, research directions: DICT and 5G.

Author unit:

Zhejiang Mobile Information System Integration Co., Ltd.

China Mobile Communications Group Zhejiang Co., Ltd. Jiaxing Branch

China Mobile Communications Group Zhejiang Co., Ltd. Hangzhou Branch

Anterior

After human society, after the farming era, industry, and information age, it is currently entering the era of intelligent interconnection and Wanzhi interconnection. Cloud computing, big data, artificial intelligence, the Internet of Things, 5G mobile Internet and computing power network have a profound impact on the operating model and production model of traditional industries.

In terms of definition, computing power is the computing power of the device through processing data to achieve the output of specific results. The core of computing power is CPU, GPU, FPGA, ASIC and other types of computing chips, and is carried by computers, servers, high -performance computing clusters, and various smart terminals. Power processing and calculation. The larger the computing power value, the stronger the comprehensive computing power. The commonly used metering unit is the number of floating -point calculations per second (FLOPS, 1 ELOPS = 10^18 FLOPS). According to estimates, 1 EFLOPS is about 5 Tianhe 2A supercomputers, or 250,000 mainstream dual -road servers, or 2 million mainstream laptops. [1]

From a functional point of view, computing power networks are distributed new computing power resource solutions with traditional centralized computing power resources represented by cloud computing to edge computing. [2] Mainly solve the problem of imbalanced computing power supply, promote the flow of computing power, change from local resources to global resources, promote the development of computing power transactions, regulate the relationship between supply and demand through the market, and provide more sources of computing power for the demanders. And choice to promote application development.

01

Overview of computing power network

Based on the new stage of development, implement new development concepts and build a new development pattern. China Mobile's computing power network construction is the basis for supporting the national network power, digital China, and smart society strategy. It is an important support for the deployment of national planning and implementation of the "East Digital and West Calculation" project. The new path will effectively promote the sustainable and healthy development of the digital economy. China Mobile is building a new development plan with 5G, computing power network, and smart China -Taiwan integration to create a new type of information service system of "connection + computing power + ability".

China Unicom released the "China Unicom Calculation Power Network White Paper" [3]. On the basis of the continuous development of computing power, basic resources such as computing and storage through the network can be effectively allocated between the cloud -edge -end. Methods to improve business service quality and user service experience.

From the layout of the operator's computing power network, the computing power network is mainly based on the network means, and the coordinated scheduling of resources (networks, computing, storage) to achieve the optimized and utilization of resources, and provide users with a consistent business experience. It aims to achieve the optimal user experience, the best computing resource utilization rate, and the optimal network efficiency. In terms of technical form, the operator plans to return to the main channel of the communication pipeline, and develops the digital transformation of business around the advantage resources. Compared with large cloud service providers and operators' computing power network services, the data center network, cloud special network, and access layer can be ended to end network resources, which are coordinated with its data center cloud resources.

02

Overview of the development of media technology industry

2.1 The main trend of media technology industry

From the perspective of the development of the media industry, through the layout of the edge intelligence, you can re -obtain the professional status of scarce content production. At the same time, the content service based on user data can be distributed on demand. Scenario perception under edge intelligence can also enhance its user insight and experience. , To promote the media industry to achieve accurate positioning in the new value network and achieve business model innovation. [4] Throughout the past 20 years, the development of computing power, network, and storage direction has leap in qualitatively, but the horizontal contrast is not difficult to find that the three are not equivalent. The network, from the previous mainstream 100MB to the current 10GBE, even entered 25 GBE and 40 GBE, surpassing more than 100 times; storage, although the performance improvement of the HDD hard disk has no big leap in the network, thanks to the iteration of the SDD process and process. The storage field has also made quite significant progress. Back to the computing power, even if the processor process has entered 7 nanometers, the increase in computing power is far less than the leap of the previous two. How traditional media responded to the opportunities and challenges brought about by the technological revolution in such a diverse and changeable media pattern, and give full play to the advantages of traditional media. [5] The computing power network needs to have the characteristics of physical fusion, logical fusion, and heterogeneous integration. It can solve the problem of uneven supply power supply, promote the flow of computing power, and change from local resources to global resources.

2.2 The pain points of the media technology industry

Most of the businesses are in static deployment, resources cannot be reused, and there are waste of idle resources. Computing resource nodes lack the ability to interact with interactive scheduling, when computing resources are overloaded, or even when unavailable, the business experience declines. [6]

In business demand, there are more and more movie effects, and the amount of calculation is increasing. When a large blockbuster may be tens of millions of machines, it is equivalent to a computer for 10,000 years. Run for 2 years. Taking "Xianxian" as an example, Li Yan, the visual effect of this film, said that the film produced 1,300 special effects lenses, and the special effects lens basically occupied about one -third of the entire film. Graval settlement combines three -dimensional environment to create a large scene. The amount of materials involved in each production project is also very large, and the requirements for the computing capabilities, drawing capabilities and storage performance of the hardware are extremely superb. High -quality special effects.

In the network environment, the rendering capacity of independent CPUs and GPUs is very limited, so network rendering can give full play to the advantages of network resources, and make full use of computing capabilities to rendering nodes to further improve the efficiency of rendering. In the innovation case, the first stage of the first stage of the 4K live encoder of the built -in 5G module was completed in Shanghai Huawei X Labs, and the 5G network communication services were opened to the end. This is the world's first integrated test to adopt Huawei MH5000 module equipment. It is also the world's first pre -commercial product that supports NSA/SA dual -modeling. It marks another step forward in the commercial process of 5G industry applications. In the future, in the "5G+4K Backpack" live method, the hardware devices in the backpack will not see CPE or mobile phone except the encoder and battery. In the mode of dynamic distribution of traffic, the 4K 4K video code flow is uploaded to the cloud video server with the help of the 5G network, and then distributed to the relevant video processing equipment. At the same time, cloud video servers also have SLA monitoring functions. In real time, information on the network KPI, business KPI, and high -quality recovery of the required network capacity security zone with high quality. In the above -mentioned innovation cases, the main concentrated mobile network is based on the fixed network in many industries. How to achieve fixed and mobile fusion, and providing customers with a fixed mobile fusion industry upgrade solution, there are still many issues to be discussed. For example, problems such as heterogeneous terminal drivers and extensive compatibility matching are needed to meet the rapid innovation of cloud border fusion media AI applications, unified management of platforms that support edge equipment, and reduce management expenses; Edge equipment, reduce equipment costs and development costs.

2.3 Media Technology Industry Network Demand

At present, many Internet companies have proposed cost reduction and efficiency, including media technology companies. Talent reduction has become a very important KPI for many companies CTOs, and it has also become an inevitable trend of technological development. The elastic service of the computing power network is a way for many companies to use extensive cost reduction and efficiency.

To determine whether a technology is a future trend, it is necessary to comprehensively calculate, store and the development of the network. At this stage, due to the failure of Moore's law, the performance of the CPU performance is weak, and relatively speaking, the network and storage performance have improved faster in recent years. Therefore, the enhancement of the network and storage capacity has made up for the performance losses caused by the separation of depository.

Take video collection in the media industry as an example. The previous process was to collect the video first, and then uploaded to the central cloud or local server for processing, but this can no longer meet the needs of enterprises. Now, after the collection, the company can process these video data nearby without uploading it to the central cloud to meet the needs of network delay and reduce the cost of network transmission. The network transitions from "connection computing power" to "scheduling computing power" and "organizational computing power". The integration of the network, the network -centric multi -dimensional resource integration supply, provides the optimal resource allocation. The integration of resources to achieve the optimization and utilization of the entire network of resources. In this scenario, we need to compare the computing power and network perception of network perception, so that the edge container can manage computing power management, including the sinking of the algorithm. 03

Computing power network and media industry integration development strategy

3.1 Industrial demand and technological evolution

Promote business development with innovation theory. The results of the study show that there is no corresponding relationship between disruptive technology innovation and emerging industries. [7] The computing power network is currently in the stage of coordination network, which mainly solves the problem of one -stop service and collaborative operations. Later, follow the integration service, unified operation and integration service, and innovate model innovation.

3.2 Commercial foundation and key technologies

Support technical practice with business innovation. For heterogeneous acceleration scenarios, the implementation architecture of heterogeneous acceleration is usually the CPU+GPU/FPGA/DSA. It is mainly calculated by the CPU to complete the non -accelerated part and the control scheduling of the entire system. The GPU/FPGA completes the acceleration of specific tasks. In terms of network perception of key technologies, we need to change the real -time linkage of network reconciliation, and the real -time synchronization of the computing power deployment status of cloud adjustment is required. Switching of computing power nodes that are provided quickly and triggered by the network, the services and data of two edge nodes are required to synchronize quickly. In terms of network perception of key technologies, the media application ecology needs to comply with the requirements of the network allocation of information such as the network, and the terminal needs to support the information of the app-ID in the message to the network APN6 and the edge applications. Infrastructure service support.

04

Development suggestions for the integration of computing power network and media industry

4.1 Lightweight of the service carrier

With the continuous development of software definition storage, more and more open source distributed databases have stabilized the heel in the field of special data access. For calculating tasks with a huge amount of calculation, you may need to divide the computing tasks, and uninstall each sub -task to at most edge computing nodes. How to divide and distribute the computing tasks efficiently is also an important issue for the optimization of energy consumption optimization of computing power networks. Essence [8]

4.2 Business and computing power network collaboration

The main line through the "data + computing power (computing network / algorithm) + industry scenario", insist that the data only has the value of value after the industry scenario is superimposed. Accelerate the level of computing power computing power, strengthen digital transformation support capabilities, and promote public computing power to generally apply. 5.5G (ultimate wireless connection), F5G (ultimate wired connection), IPv6+(certain IP), Yunhua AI (intelligent computing), quantum computing (high performance calculation), blockchain (security) and other new technologies gradually mature , Support the implementation of the operator's computing power network planning. In response to heterogeneous composite resources, build a unified computing force and modeling system, form a unified measurement criterion for computing power, realize the synergy of heterogeneity, and support the optimization of basic resources and computing power operations.

4.3 Innovation of business model

Amazon is a giant in the field of cloud computing today. Among them, the auction cloud service has been welcomed by users for its ease of use and low prices. Therefore, studying its pricing model can help cloud service providers improve their pricing methods and obtain more profits. On the other hand, help users choose the payment model that suits them and save costs. [9] There are two standards from "seeing the way" to "Seeing it clearly". The first criterion is that you can draw a picture, for example A formula, this formula is your successful formula. Successful equations follow the principle of winning more.

According to the weight value of the logical door preset, generate the computing power of the initial logic circuit; according to the computing power, receive the corresponding pre -deducted digital assets of the demand party; according to the initial logic circuit and input data to generate the calculation task, the calculation task is sent to the outsourcing calculation device ; Receive and verify the calculation results sent by the outsourcing calculation device; when the calculation results are correct, send the pre -deducted digital assets of the demand party to the outsourcing calculation device. [10]

05

Conclusion

The establishment of a new development pattern must be clearly clarified, which can only be found from the new stage of development and new development concepts. [11] From the perspective of network development, the development concept of computing power network is a new type of network system for computing and intelligent services. Important direction of network evolution. The goal of the computing power network is to provide the integration network service. It is a new stage of the integration service of the cloud network, and it is a new form of digital infrastructure services. Calculate the computing power information, comprehensive computing power and network information through the network protocol, provide the best computing power distribution and network connection for traditional media applications, and flexibly dispatch the traditional media data to the optimal computing power node to achieve traditional traditional computing power nodes to achieve traditional traditional computing power nodes to achieve traditional tradition Efficient connection between media data and computing power. //references

[1] China Information and Communication Research Institute. China Computing Power Development Index White Paper [R]. 2021.

[2] Lu Tingjie, Liu Feng, in the context of the digital economy [J]. Journal of Beijing Jiaotong University (Social Science Edition), 2021 (1): 8.

[3] China Unicom Network Technology Research Institute. China Unicom Calcum Power Network White Paper [R] .2019.

[4] Yu Guoming, Li Fengping. Media innovation in the 5G era: Application paradigm on edge computing [J]. Journal of Shanxi University (Philosophy and Social Science Edition) 2020 (1): 65-69.

[5] Chen Xu Guan. Sharing Party News Platform Resources Talk to the Media Integration Path "Integration Development Path and Vision: Ask Regional Media-" China Media Technology "Media Salon Event and Yunnan Provincial Party Newspaper Financial Media Platform Cooperation Framework Agreement Signing Ceremony" A [J]. China Media Technology, 2017 (1): 16.

[6] Lei Bo, Wang Jianglong, Zhao Qianying, Yu Yongzhi, Yang Mingchuan. Based on computing, storage, and transmission of new network virtualization architecture [J]. Telecom Science, 2020 (7): 42-54.

[7] Zhu Chengliang. The interactive mechanism of subversive technology innovation and industrial development-dual perspective based on supply side and demand side [J]. Inner Mongolia Social Science, 2020 (1): 112-117.

[8] Jia Qingmin, Ding Rui, Liu Hui, Rong Guoping. Research on the research progress of computing power network [J]. Journal of Network and Information Security, 2021 (5): 1-12.

[9] Analysis of Li Xuefei, Li Yan, Zhang He, and other. Analysis of the pricing strategy of Amazon Bidding Cloud Service [J]. Small micro-computer system, 2019 (6): 1236-1241.

[10] Li Shenglin, Sun Lilin. A computing power trading method and computing channel, CN109064329a [P]. 2018.

[11] The basic principles, challenges and policies focusing on the basic principles of the new development pattern of Deng Quheng, Liu Hong, and Wang Qiong. [J]. Learning and exploration, 2021 (12): 103-112.

- END -