Shulou(Shulou.com)06/03 Report--

KubeEdge open source background

KubeEdge announced open source in Shanghai KubeCon on November 24, 18, and there was a wave of discussion about edge computing in the technology circle, which opened a new chapter in the linkage of edge computing and cloud computing.

KubeEdge (Kube+Edge), as its name implies, relies on the container choreography and scheduling capabilities of K8S to achieve cloud-side collaboration, computing sinking, and smooth access to massive devices. The KubeEdge architecture consists of two parts, the cloud and the edge side. The cloud is responsible for the distribution of applications and configurations, while the edge side is responsible for running edge applications and managing access devices.

Previously, the deployment of applications on the edge side with K8S has been in the experimental stage. KubeEdge was born out of Huawei Cloud IEF Service and is the first open source project in the field of edge computing with the ability to deploy in a production environment. A few days ago, the K8S IOT/Edge working group released a white paper on edge computing and used KubeEdge as the reference architecture of K8S in the IOT/Edge scenario.

Note: the content of the white paper "Container Rubik's Cube" official account will be deeply interpreted, please look forward to it!

After KubeEdge open source, there are also some open source edge computing platforms and foundations at home and abroad. But when you take a closer look at these projects, LF Edge's mix of five projects is a bit like a hodgepodge, OpenEdge strong binding vendor neutrality is debatable, and K3s changed K8S in order to deploy a complete cluster at the edge, none of them seems to be a complete open source solution from cloud to edge to device. When writing and comparing these open source edge computing projects, TheNewStack wrote:

It would appear openEdge is integrated with Baidu's IoT cloud platform so it is not clear how independent is the technology.

Unlike other open source or half-open projects, whether it is community ecological construction, project operation or code development, KubeEdge put openness first at the beginning of its birth. So, where on earth is the openness of KubeEdge? What far-reaching impact will this major version update have on the entire marginal computing industry?

V0.1-ALL in K8S, extreme optimization

The natural lightweight and portability of containers are very suitable for edge computing scenarios, which is well known to manufacturers and developers of edge computing. And in view of the fact that K8S has become the de facto standard of cloud native orchestration, joining hands with K8S to enter the edge will probably end the current chaotic state of edge computing and define a unified standard for application deployment and management in the cloud and edge.

However, due to the instability of edge scene communication and strict resource consumption restrictions, the native K8S components can not run directly on the edge nodes, such as industrial gateways. However, due to the disconnect problem caused by the list/watch mechanism of K8S, the local autonomy can not be achieved after the data surface is disconnected from the management plane.

KubeEdge chooses the "light edge" architecture, that is, the container engine and device management agent on the edge side are as lightweight as possible, the management side runs in the cloud, and is built on the scheduling capability of K8S, which is 100% compatible with K8S native API. The design philosophy of KubeEdge all in K8S allows users to customize requirements around K8S's standard API or easily integrate mature projects in the cloud native ecosystem.

It is worth mentioning that the core members of the KubeEdge project come from maintainer and TOP contributor of the K8S/CNCF community. They are well aware of the design, concept and implementation mechanism of K8S, and naturally understand the problems that K8S needs to solve in the field of edge computing. In order to enable KubeEdge to run in almost all edge scenarios (for example, industrial manufacturing, smart park, intelligent driving, etc.), and to weigh the resource consumption and native capability support of K8S, KubeEdge chooses to fully rewrite the edge implementation on the basis of docking with K8S API. Currently, the disk footprint of the KubeEdge edge component is about 60MB, and the runtime memory footprint is less than 30MB.

The open source v0.1 version of KubeEdge includes five edge-side components: edged,edgehub,eventbus,devicetwin and metamanager. The architecture is shown below:

Where:

Edged is a redeveloped lightweight Kubelet to realize the life cycle management of K8S resource objects such as Pod,Volume,Node.

Metamanager is responsible for the persistence of local metadata and is the key to the autonomy of edge nodes.

Edgehub provides reliable cloud-side information synchronization, which solves the disconnect problem caused by K8S list/watch.

Devicetwin is used to synchronize information to the cloud

Eventbus subscribes to device data from MQTT Broker

It is not difficult to find that the greatest significance of v0.1 version of KubeEdge is to bridge the gap between K8S and edge computing. It provides functions such as seamless connection to K8S ecology, data localization processing, offline autonomy of edge nodes, simplification of application deployment on the edge side, multi-scene support and so on, which provides an example for cloud native edge computing architecture.

V0.2-A milestone in ecological construction

If the v0.1 version of KubeEdge is just a brick to attract jade, then v0.2 version will be an important milestone in ecological construction. Because of its open design, KubeEdge has attracted the attention of many community developers and manufacturers as soon as it is open source.

Although the v0.1 version can enjoy the ability of synergy with Huawei Cloud IEF service, in order to enable more overseas users to fully experience KubeEdge and further dispel users' concerns about vendor binding, the KubeEdge team decided to release the cloud code in advance, that is, in the latest v0.2 version. From KubeEdge v0.1 to v0.2, although it is only a small step after the decimal point, it is a milestone in the development of edge computing. The release of KubeEdge v0.1 and v0.2 set two records respectively:

The world's first open source edge computing platform supporting container deployment; the world's first open cloud edge computing platform with collaborative capabilities

The release of KubeEdge v0.2 means that anyone can deploy a complete edge computing solution covering clouds, edges, and devices in their own environment. This update adds two important components to the cloud based on v0.1, namely, CloudHub and EdgeController. The architecture is as follows:

Where:

Cloudhub is responsible for receiving information synchronized by EdgeHub to the cloud

EdgeController is used to control the state synchronization of K8S API Server and edge nodes, applications, and configurations

Users can manage edge nodes, devices and applications in the cloud directly through the kubectl command line, and their usage habits are exactly the same as those of K8S without having to re-adapt.

Roadmap

After giving KubeEdge fully open cloud edge collaboration capabilities, we will work with cloud native communities and edge computing ecological players to gradually improve the KubeEdge project from the following aspects to build an independent and controllable edge computing open source community!

Describe the choreography information of the device through CRD, and realize the corresponding controller.

External exposure of marginal services

Support direct communication between edge nodes

Enhance the security of edge nodes

Optimize the efficiency of cloud-side message channels

Provide a general edge data management / analysis framework

Support for edge serverless

A framework for providing multi-device protocol access

……

Colored egg

The KubeEdge project is currently open source on Github, project address: https://github.com/kubeedge/kubeedge

As the first open source project of edge computing launched in China, star and PR are welcome. Any PR will be given a copy of "Cloud Native distributed Storage Cornerstone: etcd in-depth Analysis":)

For related services, please visit:

Https://support.huaweicloud.com/cce/index.html?utm_content=cce_helpcenter_2019

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.