当前位置:网站首页>Kubeedge's core philosophy
Kubeedge's core philosophy
2022-06-30 12:49:00 【Brother Aaron】
The core idea
- Cloud side collaboration
- Bidirectional multiplexing message channel , Support edge node in private network ( The edge network does not need a public network Ip, You only need to have the ability to access the public network , Support for home networking , LAN, etc )
- Websocket+ Message encapsulation , Dramatically reduce communication pressure , It can still work under high time delay
- Edge offline autonomy
- Node metadata persistence , Implement node level offline autonomy
- Node failure recovery does not require List-watch, Reduce network stress , Fast ready ( Native k8s After the failure , Need to look like master Total quantity list To restore )
- Extremely lightweight
- restructuring Kubelet Function module , Ultra lightweight (~10mb Memory footprint )
- Remove the embedded storage driver , adopt CSI Access
- Support CRI Integrate Containerd、CRI-O, Optimize runtime resource consumption
Detailed introduction :
- Cloud side collaboration Cloud side collaboration is KubeEdge A highlight of .KubeEdge adopt Kubernetes standard API Managing edge nodes in the cloud 、 Add, delete, modify and check the equipment and workload . System upgrades and application updates of edge nodes can be directly distributed from the cloud , Improve the operation and maintenance efficiency of the edge . in addition ,KubeEdge The bottom optimized multiplex message channel is relative to Kubernetes be based on HTTP Long connection list/watch The mechanism is more extensible , Allow access to a large number of edge nodes and devices .KubeEdge Cloud components are completely open source , Users can be in any public cloud / Deploy on private cloud KubeEdge And don't worry about vendor lock-in , And freely integrate other services of the public cloud .
- Edge node offline autonomy And Kubernetes The nodes of the cluster are different , Edge nodes need to work autonomously in a completely disconnected mode , There is no periodic state synchronization , Only when reconnecting will it communicate with the control surface . This mode is consistent with Kubernetes The management surface and work nodes are connected by heartbeat and list/watch The original design that keeps the status updated is very different . KubeEdge The off-line autonomy of nodes is realized by message bus and local storage of metadata . The control plane configuration and device real-time status updates expected by users are synchronized to local storage through messages , In this way, the node will not lose the management metadata even if it is restarted offline , And maintain the management ability of the node equipment and application .
- Equipment management KubeEdge It provides a pluggable unified device management framework , It allows users to develop device access drivers according to different protocols or actual needs on this framework . Currently, the agreements supported and planned to support are :MQTT,BlueTooth,OPC UA,Modbus etc. , As more and more community partners join ,KubeEdge More device communication protocols will be supported in the future .KubeEdge adopt device twins/digital twins Update and synchronize equipment status , And provide... In the cloud Kubernetes An extension of API Abstract device objects , Users can use... In the cloud kubectl operation Kubernetes Manage edge devices as resource objects .
- Lightweight In order to Kubernetes Deploy on edge ,KubeEdge A lightweight transformation has been carried out .KubeEdge Retain the Kubernetes The management of , Redeveloped lightweight nodes agent.
all 1202 Years. , Do not know edge computing and edge computing platform kubeedge Do you - Listen to the wind chanting articles - You know https://zhuanlan.zhihu.com/p/342626632
边栏推荐
- 【 surprise】 la vitesse de téléchargement de Thunderbolt n'est pas aussi rapide que celle de la machine virtuelle
- [one day learning awk] Fundamentals
- Charles打断点修改请求数据&响应数据
- Flink SQL console, group not recognized_ Concat function?
- JMeter之性能测试流程及性能测试关注点
- FFMpeg AVBufferPool 的理解与掌握
- Efficient elliptic curve point addition and multiplication in scrypt
- 【C语言深度解剖】float变量在内存中存储原理&&指针变量与“零值”比较
- 2022-06-23 帆软部分公式及sql生成(月份、季度取数)
- Tronapi-波场接口-源码无加密-可二开--附接口文档-基于ThinkPHP5封装-作者详细指导-2022年6月29日21:59:34
猜你喜欢

Dark horse notes -- wrapper class, regular expression, arrays class

Analysis of the whole process of common tilt data processing in SuperMap idesktop

Why should offline stores do new retail?

How to use AI technology to optimize the independent station customer service system? Listen to the experts!

电机控制park变换公式推导
![[yitianxue awk] regular matching](/img/a6/608ec8d0808dfae04d19dfeea66399.png)
[yitianxue awk] regular matching

电机控制Clarke(α/β)等幅值变换推导

【OpenGL】OpenGL Examples

黑马笔记---集合(Collection的常用方法与遍历方式)

Sarsa notes
随机推荐
全面解析免费及收费SSH工具的基本特性和总结
ffmpeg 杂项
2022-06-23 帆软部分公式及sql生成(月份、季度取数)
Flinksql customizes udatf to implement topn
How to use AI technology to optimize the independent station customer service system? Listen to the experts!
Unity脚本的基础语法(3)-访问游戏对象组件
SuperMap 3D SDKs_ Unity plug-in development - connect data services for SQL queries
Android development interview real question advanced version (with answer analysis)
JMeter性能测试之相关术语及性能测试通过标准
基于ThinkPHP5封装-tronapi-波场接口-源码无加密-可二开--附接口文档-作者详细指导-2022年6月30日08:45:27
两批次纯牛奶不合格?麦趣尔回应:正对产品大批量排查抽检
【一天学awk】数组的使用
Discussion on JMeter operation principle
elementui中清除tinymce富文本缓存
The realization of QT the flipping effect of QQ weather forecast window
Qt读写Excel--QXlsx工作表显示/隐藏状态设置4
排查问题的方法论(适用于任何多方合作中产生的问题排查)
Today in history: Microsoft acquires PowerPoint developers; SGI and MIPS merge
时空预测2-GCN_LSTM
解决numpy.core._exceptions.UFuncTypeError: ufunc ‘add‘ did not contain a loop with signature matchin问题