当前位置:网站首页>Ali cloud senior technology expert Bai Changming: technical challenges and application innovation of edge cloud

Ali cloud senior technology expert Bai Changming: technical challenges and application innovation of edge cloud

2020-12-07 19:16:04 Aliyun yunqi

With 5G The start of the business cycle and the development of new infrastructure , 5G+ Edge computing drives and empowers the digital industry , Gradually formed the anticipated considerable industrial scale .5G In cycle , Directly and indirectly, the industrial scale has reached a trillion level , In such a huge market , There will be more and more industries with the technical support of digital transformation ,5G+ How edge computing helps industries upgrade ? In recent days, , Bai Changming, a senior technical expert from Alibaba cloud, was invited to attend 2020WICC Communication cloud conference released 《5G Edge of computing : Technological challenges and application innovation of edge cloud 》 keynote speech , analyse 5G The evolution of cloud technology on the edge of the times .

 Bai Changming WICC Photos of the speech .JPG

Where does the driving force of edge computing come from ?

Bai Changming believes that it is driven by the needs of the scene . stay 5G Previous industry application scenarios , Or through cloud computing services , Or it's a simple network service . and 5G The beginning of network construction , It is formulated for the needs of industrial Internet , therefore , At present , Traditional consumption scene upgrading and digital transformation and upgrading of the industrial scene , Need to be 5G The Internet + Cloud computing cloud network convergence technology to provide services , And edge computing is exactly the best focus of cloud network convergence , This leads to consumption and industry scenarios , The status quo of the development of two wheel drive edge computing .

The reason why edge computing can be 5G The focus of the new scene , Compared with the traditional centralized public cloud , Edge cloud computing has a large number of nodes 、 Small scale of single point 、 Technical features of distributed cloud computing close to the terminal , And these technical features are in line with 5G The low latency of the network 、 Large bandwidth 、 Large connection and localization technology features are well matched and integrated . Only by truly integrating cloud and network can we meet the new requirements of industrial Internet application scenarios .

 Screenshot  2020-11-02  In the morning 10.45.44.png

With 5G Integration and development of edge computing technology , The new technology itself also has a great impact on the data production and consumption patterns , future , Most of the 5G Business generated data and computing requirements will happen on the edge side . The most typical demand is that massive terminals of the Internet of things will generate massive links and data in the future , If the connection and data of large-scale terminals are sent back to the central cloud for processing at the same time , There will be great challenges to bandwidth and latency on the cloud . that , Through the distribution of large-scale distributed edge nodes closer to the terminal , The terminal data and connection can be processed nearby , Only persistent storage , On a large scale AI Key calculations such as training are sent back to the cloud Center for processing , In this way, we can meet the needs of new large-scale and high concurrency data processing scenarios .

Edge cloud technology and 5G The technical challenges of cloud network convergence

stay 5G+ There are many technical difficulties and challenges in the development and practice of edge computing , First of all, from the perspective of edge cloud technology , Because the edge cloud is a large scale 、 Distributed 、 Cloud computing of heterogeneous resources , Compared with the traditional centralized public cloud model , Will face “ Poor flexibility ”、“ Operation and maintenance is very difficult ”、“ Security / Lack of reliability ” and “ Heavy assets / The high cost ” Other questions .

 Screenshot  2020-11-02  In the morning 10.45.55.png

How can we solve the problems caused by the above edge cloud technology difficulties ? by way of practice , Alicloud has summed up some feasible technical solutions . for example , Through the instance between different edge nodes “ Dynamic migration ”, To increase the resources on the edge cloud “ Stretch and stretch ” Ability , adopt “ One click release ” It can quickly deploy and install application instances in a distributed way , Through the hierarchical management and control scheme to achieve in the case of network interruption “ Marginal autonomy ”, And provide application scenarios with cloud native technology severless And so on , In order to reduce the cost of data and computing on the edge cloud for customers . Final , And through “ Standard interface ” In the form of , From the edge of cloud technology architecture level , The external experience is consistent with the industry , Reduce the technology cost of industry chain between upstream and downstream of edge computing industry .

Again from 5G What are the challenges from a network perspective .5G The network is a global standard network , The ability to overlay cloud computing over a standard network , This forms the cloud network integration technology , And this is also the development of the network , It's a very big technical challenge . The Internet is compared to the cloud , Standards are clear , Location deployment and network coverage certainty is strong , And the edge cloud is deploying , It is flexible in form and other aspects , How to integrate deterministic network and flexible cloud computing , There are many technical difficulties to be solved . such as , How to set up 5G Network and cloud computing direct network collaboration , Resource coordination , Control and collaborative configuration , Even to the billing , Operation and maintenance , Security and other collaborative issues , You have to find some trade-offs between the cloud and the web , Innovative methods and processes .

Standardization of edge cloud computing

What means can solve some of these problems ? Bai Changming thinks that standardization is a feasible way of thinking .

In standard based 5G On the Internet , Make some technical standardization for cloud computing and cloud network integration , We can define and standardize the system of edge computing at the architecture level , such , Within the framework of standards , Using flexible cloud computing technology , It is relatively easy to play the advantages of cloud network , It also provides necessary and flexible development space for edge cloud computing .

Alibaba cloud has always been at the forefront of the industry in the development of edge cloud standards , from 2018 Released in 《 Edge cloud computing technology and standardization white paper 》, The technical definition and concept of edge cloud are defined for the first time , After that, he took the lead in compiling the first standard of general technical requirements for edge cloud in China , And edge cloud service trust ability requirements , It is also the first group of edge cloud service providers that have passed the edge cloud standard compliance test certification . at present , To meet the new technology requirements of cloud network convergence in the future , Alicloud's new architecture and edge are being developed .

 Screenshot  2020-11-02  In the morning 11.18.01.png

5G+ Innovative application scenario practice of edge computing

Next , Bai Changming shared Alibaba cloud's experience in 5G+ Some innovative scenarios in the field of edge computing .

One 、 City brain

Alibaba cloud is the leader of brain technology in domestic cities , Also in many cities in China, there are cases and solutions of urban brain . Edge computing can be used to process the video stream of urban road network camera in urban brain .

 Screenshot  2020-11-02  In the morning 10.46.04.png

such as , In the high-speed cloud scene , be based on ENS Complete high-speed video on the cloud , You can use it as soon as you have it 、 Take current nearby 、 Fast access features . By gathering the camera data at the edge and AI analysis , Then the structured data is sent back to the central cloud for business processing , Smart high speed will get more real-time 、 A more economical experience .

Two 、 HD live

The business characteristic of HD live broadcasting is that it takes up a lot of resources and bandwidth , The pressure of big data back to the cloud , therefore , Computing node resources through the widely distributed edge , You can handle HD video at the edge , codec , Save back to cloud bandwidth pressure , At the same time, the content can be distributed and transmitted directly in the edge network , Improve the efficiency of data transmission .

 Screenshot  2020-11-02  In the morning 10.46.17.png

3、 ... and 、 Intelligent transportation - Car networking

At present, online car Hailing has high requirements in terms of travel and passenger safety , therefore , A large number of online car Hailing are equipped with in car cameras , It's used to monitor the inside and outside of the car in real time , In order to protect the safety of passengers and drivers . And the car is a big mobile terminal , It's a typical wireless network transmission scenario , This needs to go through 5G The network transmits the on-board data in the car , coordination ENS Edge cloud computing nodes , Localized operation of vehicles and localization of data . It can reduce a lot of costs for the transportation industry .

Four 、 edge AI

Conventional AI Calculation , After big data and model training on the cloud , Direct reasoning , Or run reasoning directly on the terminal device . But because of the volume of the terminal device , Calculate the force , Power waste , Temperature and other conditions , quite a lot AI Computing can't be done directly on the terminal , in addition , The iterative speed of software update is faster than that of terminal hardware , This is right AI Algorithm and AI Chip matching also presents great challenges . Edge nodes are typical cloud computing nodes , It's for AI Training for , Reasoning , Model iterations provide good resources . future , Large scale data model training is still done by the central cloud , The reasoning can be carried out from the center to the edge node nearest to the terminal , It not only meets the requirement of low delay , At the same time, it also meets the requirements of cloud terminal AI Computational demand .

 Screenshot  2020-11-02  In the morning 10.46.29.png

“ With , Practice of edge computing scenarios , We sum up the findings , At present, the scenes and industries where edge computing is relatively concentrated are all focused on the flow of people 、 terminal 、 The city area with dense IOT equipment produces , For example, vehicle road coordination in road computing , Traffic monitoring . For example, cloud games in home computing ,AR/VR etc. . therefore , Alibaba cloud also proposed the concept of city computing and scene computing this year , Through innovative industry applications, edge computing has a more clear and clear positioning . To help the industry better 、 It's easier to understand and use the new technology of edge computing .”

Create a scene based cloud edge collaborative open Ecology

Alibaba cloud's ecological layout in the field of edge computing , Vertically , Based on IDC,MEC, A large number of heterogeneous edge infrastructures such as client side nodes , In this layer , Through the model of open cooperation , We have built diversified edge nodes all over the world . Above the computing resources is the operating system layer , Ali based on 10 Years of accumulation of flying operating system , Self developed edge oriented edge cloud operating system , Combined with AliOS This kind of end-to-side operating system , A distributed collaborative computing platform based on experience is constructed . Above the operating system layer is the capability open layer ,5G+ The innovation scene on the edge is not just the use of edge resources , At the same time, it also needs a lot of capability middleware to provide services , Scene oriented , Built with CDN cache , Video codec ,RTC,IOT,AI Algorithm ,5G Open layer that can open equal ability , Provide middleware capabilities with edge attributes for scenarios . At the top is the ecological application layer , At present, Alibaba cloud's edge computing is not only the e-commerce in Alibaba ecosystem , Finance , logistics , Entertainment and other businesses to provide services , At the same time, it also provides services for the ecological application of the whole industry .

Bai Changming says :“ The edge of Alibaba cloud , It is to create a scene based cloud edge collaborative open ecology , Welcome the industry partners to cooperate and win together , Build together 5G Edge computing new ecology .”

Link to the original text
This article is the original content of Alibaba cloud , No reprint without permission .

版权声明
本文为[Aliyun yunqi]所创,转载请带上原文链接,感谢
https://chowdera.com/2020/11/202011101041184122.html