Introduction: KubeEdge is an open source edge computing platform. Based on the original container arrangement and scheduling capabilities of Kubernetes, it extends to realize cloud-side collaboration, computing sinking, massive edge device management, edge autonomy and other capabilities. KubeEdge will also support scenarios such as 5G MEC and AI cloud-side collaboration in the form of plug-ins, which have been applied in many fields.

This article with harmonic cloud technology in China Mobile and a cross-sea bridge terminal equipment monitoring data acquisition practice case, to analyze how to land edge computing.

Current status of edge computing ecology

As of The end of November, there were 3.9 million results for Edge Computing on Google Scholar, compared to 1.3 million in October.Gartner, the world’s leading information technology research and consulting firm, annually forecasts the top 10 strategic technology trends for the coming year. Edge computing has been listed as one of the top ten strategic technology trends for the future by Gartner for four consecutive years since 2018.

The open source community currently has three main frameworks. KubeEdge is an edge computing scenario, designed for collaborative design of cloud computing framework the edge of the industry’s first cloud native, on Kubernetes native ability of container arrangement scheduling application of synergy between the edge of cloud, resource synergy, such as data together and equipment coordination ability, full through the cloud in the calculation of edge, edge, equipment together.

Edge computing scenario classification

As shown in the figure:

How do edge calculations land?

Case 1: China Mobile Online cloud-side collaboration platform

Case 1 is a typical High rate scenario, mainly focusing on clouds and edges. This case has been selected into the “Distributed Cloud and Cloud Edge Collaboration Top ten Practice Case Set (2020)” released by THE Ict Institute.

Case Background:

At present, we have cooperated for three consecutive years

Created the world’s largest containerized call platform

In 2018-2019, construction of cloud center: all business will be containerized and deployed in K8s cluster

In 2019-2020, cloud-side collaborative construction will be carried out

We have built a central cloud in 2018-2019, but like 10086, there are dozens of branches across the country, and each branch has many machines, the least may be more than 10, and the most may be dozens or even hundreds of machines. At that time, the problem was that these branches also had to be containerized, through the cloud center of the head office to do the unified release and upgrade of these applications. In other words, the digital transformation needs to be extended from the cloud center to each branch company. Some business structures of the branch company use a unified CRCD release process, and each branch company handles its own computing. Some computing that was previously unified in the cloud center can be lowered into the computer room of each branch company.

We did a lot of exploring. Consider the following two basic architectures: one is to build A K8s cluster in each branch and integrate it into the K8s cluster in the cloud center; the other is to integrate the machines in each branch as a node into the cloud center for unified management.

If the first way is used, there are some defects, such as 10086, each branch may have as few as 10 machines, if you want to build a high availability K8s cluster, you need three nodes master, then its resource utilization is very low.

On the other hand, even if k3S is used for multi-cluster management, its operation and maintenance is no less difficult than K8s. Therefore, we have been looking for other solutions. At the end of 2018, KubeEdge’s open source solution was discovered and researched, eventually adopting the form of a second architecture.

Cloud side Collaboration Platform:

Central cloud is a cloud center built on the edge computing cloud platform based on harmonic cloud, including highly available K8S cluster, Master and CloudCore. Join in the cloud center as edgecore in each branch, and make unified resource scheduling in the cloud center.

From the cloud center to each branch, traditional operation and maintenance would be troublesome, but now the cloud center can use tunnels to remotely operate and maintain branch nodes, flexible expansion of cloud edge, and regional management of cloud edge.

Monitoring was done using the now common Prometheus federated cluster. We also made a lot of adaptations for fault isolation.

Capabilities of China Mobile online cloud side collaboration platform

Cross-region node access: Expand the management boundaries of the central cloud and connect edge computing resources to the central cloud

Edge autonomy: Applications can run sustainably and provide services for cloud-side weak network scenarios

Resource reuse: Based on star topology architecture, old servers of branch companies are reused to jointly create computing resource pools

Technology upgrade: assisted the transformation and upgrading of application deployment operation and maintenance mode of branch company

Cost reduction and efficiency increase: With the advanced CI/CD pipeline of container cloud standard delivery platform, the application operation and maintenance efficiency of branch company has been increased by more than 35%

Computing sink: Computing adjacent data sources to reduce data transmission costs and central cloud storage costs

Potential Medium Platform: explore the construction of video service medium platform and distributed call customer service system in 5G scenario

Final result

Case 2: Monitoring data collection of terminal equipment of a cross-sea bridge

The bridge with many terminals, such as to collect temperature humidity, be rain and snow (add water in it, immediately can display), used to collect dust can more smooth sense of PM2.5 and other equipment, the equipment requires a carrier to access the board above, we board is based on single point system, So we ended up using the board as an edgecore to access the entire edge computing cloud platform.

The edge computing box is integrated with 5G modules. After uniformly collecting data, it will be centralized to the Beidou monitoring station.

Overall Architecture

In addition to the above mentioned devices to connect some rain and snow bias generators, thermometer shelters and IMU can also connect some lost data. After collecting these data, some of them can be reported directly, and some of them can be processed by MQTT Broker through a layer of data processing, so that the data can be reported and converted, and then put into the cloud database for processing. The whole edge cloud platform provides device access and data visualization, edge resource management digital large screen, and we are also doing some edge pressing work.

conclusion

KubeEdge is committed to expanding The capabilities of Kubernetes to the edge. In September 2020, KubeEdge was promoted to CNCF Incubation program. So far, we have 3.3K stars, 900+ forks, 500+ contributors, and three community SIG. We hope to jointly promote the innovative development of cloud-side collaborative applications. I hope you can actively participate in the KubeEdge community, contribute wisdom and strength.

Attached: KubeEdge community contribution and technical exchange address

Website: kubeedge. IO

Github address: github.com/kubeedge/ku…

Slack: kubeedge.slack.com

Mailing list: groups.google.com/forum/#! The for…

Weekly community meeting: zoom.us/j/416723730…

Twitter: twitter.com/KubeEdge

The document address: docs. Kubeedge. IO/en/latest /

Welcome to the cloud native team, we will provide you with:

We update the cloud native technology dynamics, practical progress, application cases and so on every day;

Join the group and the industry technology champion, 10,000 + cloud native lovers link to learn together;

Irregularly invite cloud OG level technology to share technical combat

…………………