As the heat of edge computing rises, can its power compete with cloud computing?

With the ubiquitous development of the interconnection of all things, in recent years, the heat of Edge Computing continues to rise, and there is a strong resistance to cloud computing.
IDC predicts that more than 50 billion terminals and devices will be connected worldwide by 2020, and more than 40% of the data will be analyzed, processed and stored on the edge of the network.
So, what kind of edge computing technology is suitable for network edge?
Open data shows that edge computing is an open platform that integrates core competencies such as network, computing, storage, application and so on. The basic idea is to migrate the cloud computing platform to the edge of the network, and try to integrate the traditional mobile communication network, the Internet and the Internet of Things, so as to reduce the end-to-end delay in service delivery.
Since the development of edge computing, there have been three widely recognized technology architectures, namely MEC (multi-access edge computing), cloud computing and fog computing. Many times, people will confuse these concepts. Now we will analyze these edge calculations in detail. 

MEC, multi-access edge computing, was first proposed by the European Telecommunications Standardization Association (ETSI) in 2014. ETSI's initial members include HP, Vodafone, Huawei, Nokia, Intel and Viavi. MEC is an IT service environment and cloud computing capability for mobile network edge. It can offset the delay associated with backhaul by caching, data transmission and computing at mobile network edge, and finally realize millisecond application. MEC Basic Architecture Defined by ETSI.
Macroscopically speaking, the different functional entities in the basic architecture of MEC can be divided into three levels: network layer (Networks Level), mobile Edge Host Level (Mobile Edge Host Level) and mobile Edge System Level (Mobile Edge System Level). Previously, Leifeng Network has detailed analysis of the technical architecture of MEC, this time no more.
The application scenarios of MEC mainly include the following aspects:
Local edge services related to network connectivity and network capacity opening. For example, mobile virtual private network instead of enterprise Wi-Fi network, indoor location based on wireless network and combined with room division;
Edge proximity saves the return bandwidth and reduces the delay of video edge service. For example, edge caching combined with CDN, edge storage and recognition analysis for video surveillance;
Terminal-oriented computing migration reduces terminal cost of edge-aided computing services. For example, providing edge cloud rendering for AR, VR and games.
In addition, MEC can help videos and IT applications in intelligent parks to reduce traffic roundabout, reduce transmission delay and provide a safer data processing environment. In the field of automatic driving, MEC platform provides high-precision map, vision sharing, intelligent analysis and continuous switching functions for motorcades with the help of 5G technology to assist automatic driving. To provide more accurate, safer and zero interruption driving experience.
The application scenario of MEC is very extensive. In addition to these mentioned above, there are more vertical applications of the Internet of Things waiting for MEC researchers to unlock.
At present, three major domestic operators have launched MEC pilot deployment, including LTE Mobile Virtual Private Network, Vehicle Networking, Edge Caching, Indoor Location and so on. According to China Unicom's White Paper on Edge Computing Technology, China Unicom's 5G network MEC deployment planning, MEC deployment location and business scenarios are closely related, MEC deployment can be divided into wireless access cloud, edge cloud or convergent cloud.
Generally speaking, for uRLLC low-latency scenarios, MEC needs to be deployed in the wireless access cloud near the base station side; for eMBB scenarios, MEC can be deployed in the edge cloud in the high-traffic hot areas; for mMTC scenarios, MEC is deployed in the high-location convergent cloud, which can cover the business needs of larger areas.
In the form of wireless access cloud, MEC is deployed in wireless access cloud with base station CU unit and core network forwarding surface UPF. Local services are deployed on the base station side to provide users with shorter service delay.
Among them, the CU unit includes RRC and PDCP layers, the DU unit includes RLC, MAC and PHY layers, and the 4,000 eNB and 5,000 gNB CU units can be combined. In this way, MEC service coverage is small, and it is suitable for low-speed or even non-mobile but time-sensitive business, such as business related to sports venues, scenic spots.

Micro cloud
Compared with MEC, if MEC emphasizes the concept of "edge", micro-clouds focus more on the concept of movement.
Microcloud is the result of the Open Edge Computing (OEC) project, which was initially initiated by Carnegie Mellon University in the United States, and has since been supported by Huawei, Intel and Vodafone. Open data shows that it is an edge computing architecture that combines mobile computing platform with cloud computing, representing the middle layer of "mobile terminal-micro cloud-cloud" three-tier architecture. It is located between mobile terminal and cloud platform, and is a small mobile data center deployed on the edge of the network.
Although the micro-cloud itself is located on the edge of the network, and even closer to the user intuitively, it is mainly used to enhance mobility in similar vehicle network scenarios. It can provide abundant computing resources for mobile devices and even run directly on aircraft and vehicles. Microcloud, designed to deploy the cloud closer to the user, can be understood as a lightweight MEC.
As far as the location of micro-cloud deployment is concerned, the distance between micro-cloud and end-user is a one-hop wireless connection, such as deploying on cellular network base station or Wi-Fi base station, which provides low latency response for end-user's computing tasks. When multiple micro-clouds construct a distributed mobile edge computing environment to expand the available resources of users, load balancing of resources can be achieved by providing a dynamic migration mechanism similar to cloud platform.
Microclouds are essentially clouds, but they differ from traditional ones in the following aspects: Papid Provisioning, VM Hand-off and Cloudlet Discovery.
For example, rapid configuration, because micro-cloud is mainly designed for mobile scenarios, it will face the highly dynamic connection problem caused by the mobility of user terminals, so it must have flexible rapid configuration capabilities.

Fog calculation
The concept of fog computing was proposed by Cisco in 2012. Then, Cisco joined Arm, Dell, Intel, Microsoft and Princeton University to form the Open Fog Consortium in 2015.
Compared with MEC and micro-cloud, fog computing focuses on the application of the Internet of Things (IoT).
In February 2017, the OpenFog Reference Architecture was released by the OpenFog Consortium. It is a general technology architecture that combines seamless intelligence in the cloud with Internet of Things terminals using an open standard approach to support data-intensive requirements for Internet of Things, 5G and AI applications.
OpenFog has been transformed into a new computing model in terms of traditional closed systems and dependence on cloud computing. It is based on workload and device capability, making the calculation closer to the network edge. Fog computing distributes computing, communication, control, storage resources and services to users or distributes them on devices and systems close to users, thus expanding cloud computing to the edge of the network, which can be understood as a small cloud located on the edge of the network.
The whole fog network is composed of multiple fog nodes. The performance of a single fog node is relatively weak, but its geographical location is widely distributed. Because the nodes are geographically dispersed and do not concentrate on generating a large amount of heat, no additional cooling system is needed to reduce power consumption.
Lei Feng learned that in February this year, the Army Research Laboratory awarded Techca a $1 million contract to develop Smart Fog, a fog computing platform for battlefield soldiers. The platform trains machine learning algorithm in the case of networking, integrates data from various sources on the battlefield and processes them offline without connecting clouds, so that soldiers can apply artificial intelligence ability in the area of disconnected network. Here, it can be seen as the middle layer between the individual equipment and the cloud, allowing soldiers to access computing power and storage space at any time. 

A comparative analysis of the three
MEC, cloud computing and fog computing, as three specific models of edge computing, have many similarities and differences in deployment location, application scenario and real-time interaction. Lei Feng learned that its main manifestations are as follows:                 
In terms of deployment location, MEC is located between terminal and data center, and can be co-located with access points, base stations, traffic convergence points, gateways, etc. The deployment location of cloud and fog computing is consistent with the above mentioned MEC deployment location. In addition, micro clouds can also run directly on vehicles, aircraft and other terminals.
As far as application scenarios are concerned, MEC is mainly devoted to reducing application latency, suitable for Internet of Things, Vehicle Networking, AR/VR and other application scenarios; micro-cloud is suitable for mobile enhanced applications and Internet of Things and many other scenarios; and fog computing is mainly for Internet of Things scenarios that require distributed computing and storage.
For the mobility of the three and the real-time interactive support of the same application on different edge nodes, MEC only provides mobility management when the terminal moves from one edge node to another edge node; while micro-cloud provides the support of virtual machine image switching from one edge node to another edge node; as for fog computing, it only provides mobility management when the terminal moves from one edge node to another edge node. It fully supports communication between distributed applications of fog nodes. 

Whether MEC, cloud computing or fog computing, these edge computing have their own characteristics and applicable scenarios. According to the beginning of the article, 40% of the global data will be processed on the edge of the network. It is necessary to say that edge computing has become an important computing mode. These three edge computing modes are different types evolved through long-term development, so they are equally important for the development of the interconnection industry.
In addition, the collaboration between edge computing and cloud computing has also become the focus of attention. They can optimize and complement each other to enable the digital transformation of the industry. If cloud computing is a coordinator who is responsible for large data analysis of long-term data, edge computing pays more attention to real-time and short-term data analysis. As we know, edge computing is closer to the device side, so it provides support for cloud data acquisition and large data analysis, while cloud computing is sent to the network edge through the output instructions of large data analysis.

This article is reprinted from 物联网世界 . If you need to reprint or duplicate it, please indicate the source or source 

© 2019 - 2030 四川长虹网络科技有限公司 版权所有蜀ICP备18005700号 | 联系方式:+86-0816-2919636