DEV Community

Cover image for Kubernetes Native Edge Computing Framework, kubeEdge
Mansoor Ahmed
Mansoor Ahmed

Posted on

Kubernetes Native Edge Computing Framework, kubeEdge

Introduction
We are here to discuss Kubernetes Native Edge Computing Framework. KubeEdge is an open-source project. It was the first open-source program that Huawei makes a contribution to Cloud Native Computing Foundation (CNCF). It is as well the world’s first open edge computing platform founded on Kubernetes extension. That extension makes available cloud-side collaboration. KubeEdge’s name derives from Kube and Edge. It depends on Kubernetes container scheduling and its competencies to attain cloud-side collaboration, computational dipping, and huge device access.

KubeEdge provides a complete edge computing solution. That is based on Kubernetes with distinct cloud and edge core modules. KubeEdge is prepared to build edge computing solutions spreading the cloud. The control plane exists in the cloud. That is however scalable and extendable. The edge may work in offline mode at the same time. Similarly, it is lightweight and containerized. It can help heterogeneous hardware at the edge. KubeEdge places to save important arrangement and operation costs for edge solutions by the optimization in edge resource utilization. This creates the utmost convincing edge computing platform in the world presently, founded on Kubernetes.

Description
The main objective for KubeEdge is spreading the Kubernetes ecosystem from cloud to edge. Delivering the elementary edge computing structures it started with its v0.1. It takes along the cloud components to connect and complete the loop now with its latest release v0.2. KubeEdge allows the orchestration and organization of edge clusters like how Kubernetes accomplishes in the cloud. This unlocks unified possibilities of taking cloud computing capabilities to the edge, rapidly and powerfully.

Essential of cloud side collaboration
Edge computing and cloud computing are corresponding and not two equally limited technologies. IoT or Edge has more or fewer resemblances with cloud data centres, for example:

The edge also has the needs of computing, storing, net and further resources of the management node.
Edge applications likewise want containerization and microservices.
Edge computing hopefulness to have normal APIs and toolchains.
Safety, data or channel encryption, authentication and authorization.
The architecture and capabilities of the cloud data centre to the edge develops a matter unquestionably. At this time following are some scenarios that need cloud-side collaboration:

Preparation on the AI ​​cloud and edge implementation. The training of the AI ​​model is located in the cloud, and the implementation of the AI ​​is near to the device side.
Introducing the Microservices and DevOps to the edge would meaningfully speed up the iteration cycle of IoT software. For example, embedded devices and robots improve deployment and operation and maintenance efficiency.
Data backup junkyard. For instance, huge amounts of industrial data after encryption are kept in the cloud.
Distant control. The cloud at all sends a control signal to the edge device.
Automatic development. For now, the edge nodes are not as good as the cloud with good automatic growth capability. We can select to enlarge the load of the edge to the cloud at the top.
We have collected sufficient experience to manage resources on the cloud. The subsequent challenge at the present is how to build an edge cloud platform that spreads the management of resources on the cloud to the edge. Also, that permit us to flawlessly manage edge resources and devices. The Edge Cloud platform would emphasise the following matters:

Big or heterogeneous devices, right of entry to gateways and edge nodes.
A huge quantity of telemetry data is combined and treated for use by cloud applications.
Apparatus security and ID services.
Help remote command to the device.
Automatic formation and controlling of edge nodes and devices.
Appliance cloud-to-edge applications, deployment, and configuration.
Delivers data storage, occasion management, API management, and data examination competencies for edge application development.
KubeEdge architecture
The focal architecture principle for KubeEdge is to shape interfaces. Those interfaces are dependable with Kubernetes, be situated on the cloud side or edge side. The key subjects to be addressed by KubeEdge are:

Cloudside collaboration
Resource heterogeneity
Massive
Lightweight
Consistent device management and access experience.
KubeEdge architecture

The KubeEdge architecture is obviously distributed into three layers: cloud, edge and device layer. This is a whole open source edge cloud platform from cloud to edge to the device, reducing the fears of user vendors.

KubeEdge’s edge method entails the following five modules:

Edged is a re-established lightweight Kubelet. It gears the lifecycle organization of Kubernetes resource objects for example Pod, Volume, and Node.
Meta Manager is answerable for the resolution of local metadata. It is the main to the autonomy of edge nodes.
Edgehub is a multiplexed message channel. It supplied dependable and well-organized cloud edge information synchronization.
Devicetwin is used to intangible physical devices and make a mapping of device states in the cloud.
The event bus gives to devise data from the MQTT Broker.
KubeEdge’s cloud process contains the following two components:

Cloudhub is set up in the cloud and gets information. That information edge hub synchronizes to the cloud.
The edge controller is installed in the cloud to control the state synchronization of the Kubernetes API Server. That controls the nodes, applications, and configurations of the edge.
Kubernetes maser goes in the cloud. Through the kubectl command, the users may openly manage edge nodes, devices and applications in the cloud. The practice habits are precisely the equivalent as Kubernetes native, no essential to re-adapt.

Simple and Light
The KubeEdge Edge and Cloud core components can be set up only. They may run the user applications. The edge main has a footprint of 66MB. It’s only needing 30MB of memory to run. In the same way, the cloud core can run on any cloud node. Users may practice by running it on a laptop in addition.

The installation is simple and may be completed in few steps:

Arrange the pre-requisites Docker, Kubernetes, MQTT and openssl
Clone and Build KubeEdge Cloud and Edge
Run Cloud
Run Edge
The thorough steps for each are available at KubeEdge/kubeedge

Equipment management
KubeEdge offers a pluggable joined management framework for devices. That enables users to develop device access drivers established on different protocols or actual needs. Now sustained and planned support agreements are:

MQTT
BlueTooth
OPC UA,
Modbus, etc.
KubeEdge will support more device communication protocols in the future with increasingly community partners joining. KubeEdge updates and synchronizes device status via device twins or digital twins. That gives Kubernetes extended API abstract device object in the cloud. Users may achieve edge devices by using kubectl to operate Kubernetes resource objects in the cloud.
For more details visit: https://www.technologiesinindustry4.com/2021/08/kubernetes-native-edge-computing-framework-kubeedge.html

Discussion (0)