[Cloud Computing] From cloud computing to edge computing

1 Introduction

The concept of cloud extends from a central point and enters people's daily life and work environment. More and more software that originally ran in the cloud will run around them, improving people's lives, from medical care to transportation, entertainment, Manufacturing and more. In the future, the cloud's push to the edge will further accelerate. This article mainly summarizes edge computing products and services.

2. Overview of edge computing

2.1 Definition of edge computing

Gartner's definition of edge computing: Edge computing describes a distributed computing topology in which information processing is placed close to the things or people who produce or consume the information. Edge computing starts from the concept of mesh networks and distributed data centers, keeping traffic and processing local and decentralized in the network. The goal is to reduce latency and unnecessary traffic and establish a hub for interconnection between interested peers and data refinement for complex media types or computational loads.

Edge Computing Hype Cycle, 2019

2.2 The value of edge computing

• Real-time processing, solving problems from the source.
  Edge computing is closer to the data source and can process and analyze problems in real time with low latency. For service providers, edge computing can improve customer experience by moving applications or content to lower tiers in the network hierarchy. Additionally, they can deploy entirely new classes of services at the edge to fully respond to customer needs.

• Reduce costs
  According to related reports, by 2021, consumer-facing industries will spend more annually on edge network, computing and storage resources than core data center upgrades, making edge computing a key consideration for infrastructure vendors. factors, economies of scale will bring about cost reductions. For users or enterprises, bandwidth costs can be reduced. According to a report by Analysys Mason, most companies in vertical industries are expected to reduce costs by$10\%-30\%$ , the average saving is expected to be$10\%-20\%$。

• Edge computing enables 5G technology and opens up new use cases.
  As service providers modernize their networks, workloads and services are moving from the core network (located within the data center) to the edge of the network. Virtualization at the edge can help service providers Deploy services at the edge of the network. For example, Telefonica and Germany have joined hands with Ericsson and AWS to virtualize 5G core networks, and Google Cloud announced that it will cooperate with AT&T to create enterprise network edge use cases.

3. Edge computing service provider

3.1 CDN service provider

CDN, Content Distribution Network ( Content Delivery Network), its core value is to intelligently distribute digital content to nodes closer to users, thereby improving overall distribution efficiency, reducing network latency, and saving bandwidth resources. Its inherent edge node attributes, low Due to latency and low bandwidth, CDN itself is the prototype of edge computing .

In recent years, most CDN providers have focused their efforts on the edge to meet the new demands of real-time video streaming, security, and the Internet of Things.

3.1.1 Akamai

Akamai is a leader in delivering edge services, providing content delivery services to a large number of enterprises around the world, and has a comprehensive solution and service offering in web performance, media delivery, cloud security and network operator solutions. It is mentioned in the report that Akamai's network carrying capacity is approximately $90-At120TBps$ , Akamai delivers an average of more than 40 TBpsof web traffic per day.$40TBps\; andmore$ than 83 TBpsdelivered in March 2019 Peak traffic of$83$$TBp$$s\; .$

3.1.2 Wangsu Technology

Wangsu Technology mainly provides CDN, edge computing, cloud security, and SD-WAN products and solutions. It has ranked first in the country's commercial CDN market share for eight consecutive years. After cloud vendors entered the CDN market, the CDN market turned into a "Red Ocean" competition, and Wangsu Technology's market share continued to decline. However, domestic CDN demand accelerated.

Wangsu Technology Global Node Resources

3.2 Mobile operators

3GPP ( 3rd Generation Partnership Project3rd Generation Partnership ProjectMulti-access Edge Computing ) defines 5G edge computing and 5G MEC ( Multi-Access Edge Computing ) in the 5G system architecture . MEC adopts a distributed computing model, sinking these functions to the edge of the network, and hosting applications in highly distributed, small-scale localized edge computer rooms, close to devices and end users.

5G MEC framework

3.2.1 China Mobile

Through the 5G + edge computing capability platform, China Mobile provides edge intelligent connection and computing services at the edge of the network close to the source of things or data through 5G proprietary network access to meet the needs of the industry in agile connection, real-time business, and data processing. , application intelligence, security and privacy protection and other key needs to help the digital transformation and upgrading of the industry.

Its core products are 5G edge computing gateway and 5G edge computing platform .

3.2.2 China Telecom

China Telecom cooperates with CDN companies to deploy MEC edge CDN as an extension of the existing centralized CDN to serve multiple network users at the same time.

MEC + CDN edge content acceleration solution

3.2.3 China Unicom

In March 2020, China Unicom launched the EdgePOD edge cloud solution.

3.3.4 AT＆T

In March 2020, Google Cloud announced that it would cooperate with AT&T to create enterprise network edge use cases, including video analysis, AR/VR applications, and data-intensive sensors. AT&T will use Google Cloud's new cloud-native telecom version of the Anthos application platform, of which Anthos It is an open hybrid cloud multi-cloud application platform, including application modernization, cloud migration, hybrid cloud and multi-cloud management.

3.3 Hardware manufacturers

For hardware manufacturers, the current demand for supporting edge computing is mainly from the hardware side, such as low power consumption, powerful data processing, analysis and artificial intelligence functions, etc. Some hardware manufacturers also build software solutions, such as Intel's acquisition Smart Edge Company builds edge access platform services.

3.3.1 Intel

• Intel Core processors and Intel Atom processors
  • Intel Core processors provide the computing power and responsiveness to support many types of IoT devices, including kiosks, industrial PCs, and more. Choosing an Intel Core processor based on the Intel vPro platform enables additional remote manageability and helps improve security features.
  • Intel Atom processors are purpose-built for embedded applications, balancing performance with low power consumption in smart cameras, mobile point-of-sale devices and other edge devices where power consumption may be an issue.
• Intel Movidius Vision Processing Unit (VPU)
  • The Intel Movidius Vision Processing Unit delivers fast, low-power performance to support computer vision and artificial intelligence workloads. These are specifically designed VPUto enable parallel programmability and workload-specific hardware acceleration for deep neural network and computer vision applications in devices such as augmented or virtual reality systems and smart cameras.
• Intel Xeon D Processor
  • Edge servers bring powerful data processing, analysis and artificial intelligence capabilities closer to the point of data generation. Intel Xeon D processors deliver the high performance and reliability needed to accelerate compute, storage, memory and networking at the edge. Select SKUs of Intel Xeon D processors offer higher temperature ratings for embedded and rugged applications with integrated Ethernet and AI acceleration.
• Intel Security Technology
  • Edge-based devices and systems must keep proprietary, sensitive and personal information secure. In industries such as healthcare, data sovereignty and compliance are particularly important.
  • Intel products provide advanced security by providing built-in, silicon-enabled security technologies that help protect potential attack surfaces. Embedded in the silicon, our security technologies help create a trusted computing foundation that customers can rely on across our portfolio of products including edge, endpoint, data center, cloud and network. With this common security approach, deployment and integration are naturally simplified.

For hardware products, please refer to the product introduction on Intel’s official website:
https://www.intel.cn/content/www/cn/zh/edge-computing/edge-devices.html

3.3.2 Arm

In February 2020, Arm released two new processors with AI functions: ARM Arm Cortex-M55and ARM , which are chips designed Ethos-U55for Internet of Things (IoT ) terminal devices.IoT

3.4 Cloud computing service provider

For cloud vendors, edge networks are generally deployed from three aspects.

• The web goes to the edge . Cloud vendors continue to build Regions, POPs, and Local Zones, and launch dedicated line, VPN, and SD-WAN access services, with the goal of bringing the network closer to the edge.
• Devices go to the edge . Cloud vendors will sell standardized hardware to users, and edge hardware can connect edge areas and the cloud.
• Services go to the edge . Cloud vendors will launch edge computing platform services to make cloud and edge interoperability more standardized and experience consistent.

This section briefly introduces the edge products and services of cloud vendors. The following chapters will provide an in-depth analysis of the products and implementation of cloud vendors.

3.4.1 AWS

AWS IoT Edge services mainly include AWS IoT Greengrass, FreeRTOS, and AWS IoTdevice testers.

• AWS IoT Greengrass seamlessly extends AWS to edge devices so the data they generate can be manipulated locally while still using the cloud for management, analytics, and persistent storage.
• FreeRTOS is an open source real-time operating system for microcontrollers that makes it easy to write, deploy, secure, connect and manage low-power, small edge devices.

In addition, AWS has also launched product forms such as AWS Snowball, AWS Outposts, and , and is committed to making a difference at the edge.AWS Local Zones

• AWS Snowball : A small, lightweight, rugged, and secure edge computing, edge storage, and data transfer device.
• AWS Outposts : is a fully managed service that delivers the same AWS infrastructure, AWS services, APIs, and tools to virtually any data center, colocation space, or on-premises facility for a truly consistent hybrid experience.
• AWS Local Zones : A new form of AWS infrastructure deployment that places AWS compute, storage, databases, and certain other primary services closer to large population centers where AWS regions currently do not exist, or closer to industry and IT central location.

3.4.2 Azure

At Microsoft Build 2018, Microsoft grandly introduced an important component service in edge computing Azure IoT Edge. Azure IoT EdgeIt is Azure IoT Huban Azure service that is bundled with and used in combination. It mainly helps users to easily deploy analysis and business modules that originally ran in the cloud on the device side.

IoT Edge runtime

IoT Edge cloud interface

3.4.3 Google

Cloud IoT CoreIt is a fully managed service that helps users easily and securely connect, manage and extract data from devices around the world.

Build and train machine learning models in the cloud

Google AnthosIt is possible to modernize existing applications and build cloud-native applications in any environment to improve agility and save costs, where Anthos can be deployed at the edge, such as AT&T using the telecom version of Anthos to build 5G networks.

3.4.4 Alibaba Cloud

Link IoT Edge: Internet of Things edge computing is the expansion of Alibaba Cloud's capabilities at the edge. It inherits Alibaba Cloud's security, storage, computing, and artificial intelligence capabilities. It can be deployed in smart devices and computing nodes of different sizes, and is connected by defining object models. Devices with different protocols and data formats provide local computing services that are safe, reliable, low-latency, low-cost, easy to expand, and weakly dependent.

3.4.5 Tencent Cloud

Tencent Cloud IoT edge computing platform: IoT Edge Computing Platform, IECP.

Smart factory use cases

SuperEdge is an open source project jointly launched by Tencent Cloud and several vendors in the community. It is an edge container management system based on native Kubernetes. This system extends cloud-native capabilities to the edge, effectively implements cloud-to-edge management and control, and greatly simplifies the process of deploying applications from the cloud to the edge. SuperEdge provides strong support for edge-native applications.

SuperEdge framework

4. Cloud computing service provider edge computing architecture and platform

This chapter mainly analyzes the cloud computing service providers' construction of edge capabilities. As for the regions where cloud vendors expand their services, the evaluation is mainly based on the breadth of network products and infrastructure. Next, cloud vendors will be evaluated from the aspects of network, hardware, and services from cloud to edge. Perform analysis.

4.1 Analysis of AWS edge products and services

在 $2020re:Invent$ General Assembly,$Amazon.comVicePresident$ and CTOW erner Vogels Werner\$Vogels$$One\; ofDr.WernerVogels’$ predictions for technology trends in 2021 is that the push from the cloud to the edge will$further$$accelerate$$.$This next section introduces the products and capabilities AWS provides to empower the edge.

4.1.1 Network goes to the edge

AWS provides a variety of network products, such as VPN, dedicated line access, Transit Gateway, and services that support SD-WAN. For details, see the introduction to AWS Transit Gateway.

AWS Transit Gateway full network interconnection model

At the same time, AWS also provides many cloud Regions, Availability Zones, Local Zones, Wavelength Zones, and dedicated line access points. The recently launched Local Zones are closer to users.

Global Infrastructure Map: https://www.infrastructure.aws/

AWS Local Zones

In December 2019, AWS Los Angeles launched the first AWS Local Zones. Three AWS Local Zones will be launched in 2020, and 12 more will be launched in 2021, pushing the network closer to the edge of users.

AWS Local Zones are a new form of AWS infrastructure deployment that places AWS compute, storage, databases, and certain other primary services closer to large population centers where AWS currently does not exist, or closer to industry and IT centers s position.

How AWS Local Zones work

Local Zones deploy cloud services at the edge, allowing users to use cloud services nearby. It has the following advantages:

• Low latency for local end users
• Consistent AWS experience
• Flexible and scalable

4.1.2 Hardware goes to the edge

（1）AWS Outposts

AWS Outposts is a fully managed service that delivers the same AWS infrastructure, AWS services, APIs, and tools to virtually any data center, colocation space, or on-premises facility for a truly consistent hybrid experience.

How AWS Outposts work

（2）AWS Snowball

AWS Snowball belongs to the AWS Snow family and is an edge computing, data migration, and edge storage device. Snowball supports specific Amazon EC2 instance types as well as AWS Lambda functions. Users can develop and test in the AWS cloud, and then deploy applications on devices in remote locations to collect, preprocess, and ship data to AWS. Common use cases include Data migration, data transfer, image wrangling, IoT sensor stream capture and machine learning.

（3）AWS Wavelength

AWS Wavelength is an AWS infrastructure product optimized for mobile edge computing applications. Wavelength Zones are an AWS infrastructure deployment that CSPembeds AWS compute and storage services in communications service provider (CSP) data centers at the edge of the 5G network, so application traffic from 5G devices can flow without leaving the mobile communications network. Reach the application server running in the Wavelength zone. This avoids the latency caused by application traffic having to traverse multiple hops across the internet to reach its destination, allowing customers to take full advantage of the low latency and bandwidth offered by modern 5G networks.

How Wavelength works

4.1.3 Services move to the edge

（1）AWS IoT Greengrass

AWS IoT Greengrass is software that extends cloud capabilities to local devices, seamlessly extending AWS to edge devices so the data they generate can be acted upon locally while still using the cloud for management, analytics, and persistent storage.

How AWS IoT Greengrass works

（2）AWS IoT Core

AWS IoT Core is a managed cloud service that allows connected devices to easily and securely interact with cloud applications and other devices.

How AWS IoT Core works

4.2 Alibaba Cloud

4.2.1 Network goes to the edge

Alibaba Cloud provides a wide range of network products, such as VPN, dedicated line access, cloud enterprise network, cloud connection network, intelligent access gateway, and SD-WAN access services. For details, see Alibaba Cloud Enterprise Network products and implementation analysis.

4.2.2 Hardware goes to the edge

Alibaba Cloud's IoT edge computing series all-in-one machine ( LE) series is a product designed for IoT scenarios that completes calculations in the customer's local data center and provides cloud-edge remote collaboration services.

(1) Edge application all-in-one machine

(2) Edge video intelligent all-in-one machine

4.2.3 Services move to the edge

From the perspective of Alibaba Cloud's services, it mainly provides vertical products.

(1) Alibaba Cloud official website IoT products

（2）Link IoT Edge

IoT edge computing Link IoT Edge is the expansion of Alibaba Cloud's capabilities at the edge.

4.3 Tencent Cloud

4.3.1 Network goes to the edge

Tencent Cloud provides a wide range of network products, such as VPN, dedicated line access, cloud networking, and SD-WAN access services. For details, see Tencent Cloud Cloud Networking Products and Implementation Analysis.

Tencent Cloud Global Infrastructure

Tencent Cloud's expansion of network regions is currently provided by EC. EC points are deployed in second and third largest cities, such as Wuhan, Changsha, Zhengzhou, Xi'an, etc. Tencent Cloud has opened servers in six cities.

4.3.2 Hardware goes to the edge

(1) SD-WAN Edge device

Tencent Cloud's cloud networking provides Edge devices with SD-WAN access capabilities, which can be deployed in stores, highways, mobile news vans, etc. Edge can access 5G capabilities.

（2）Mini T-block

Mini T-block is designed for data centers, can run independently, is plug-and-play, and is as convenient to use as a USB flash drive. Mini T-block is lighter and requires lower space, which greatly reduces the entry threshold and investment cost of the data center.

4.3.3 Services move to the edge

(1) Tencent Cloud official website IoT products

（2）SuperEdge

SuperEdge is an edge container management system based on native Kubernetes. This system extends cloud-native capabilities to the edge, effectively implements cloud-to-edge management and control, and greatly simplifies the process of deploying applications from the cloud to the edge. SuperEdge provides strong support for edge-native applications.

5.Thinking and Outlook

• Open source : A single supplier alone cannot provide a complete edge computing solution. The solution must be the result of the joint efforts of multiple parties and needs to borrow the advantages of open source. When using open source solutions, edge technology can be applied to a wider ecosystem, and No vendor lock-in to a proprietary technology stack.

• Cloudification : Edge computing is undergoing cloudification, and edge devices and programs have the product capabilities to use public clouds to achieve greater value.

• Standardization : Edge gateways need to be deployed in multi-cloud and heterogeneous devices. Standardization is also very important and can be deployed in various complex network environments. For example, Google Anthos, Tencent's SuperEdge, etc. are built on container platforms.

• Platformization : The accelerated development of 5G provides platform-based edge computing capabilities, allowing enterprises or service providers to participate in edge computing services.