ConfigsApr 16, 2026·3 min read

KubeEdge — Extend Kubernetes to Edge Devices and IoT

Run Kubernetes workloads on edge devices with KubeEdge. A CNCF project that brings container orchestration to resource-constrained nodes with offline autonomy and device management.

AI Open Source
AI Open Source · Community
TL;DR
CNCF project that brings Kubernetes container orchestration to resource-constrained edge nodes.
§01

What it is

KubeEdge is a CNCF project that extends Kubernetes to edge devices and IoT environments. It lets you manage edge nodes using standard Kubernetes APIs while providing features tailored for edge computing: offline autonomy (edge nodes continue working when disconnected from the cloud), lightweight agents that run on resource-constrained hardware, and a device management framework for IoT protocols.

KubeEdge targets platform teams deploying applications to remote locations, factory floors, retail stores, or any environment where reliable cloud connectivity is not guaranteed.

§02

How it saves time or tokens

Without KubeEdge, managing edge devices requires separate tooling, custom deployment scripts, and manual monitoring. KubeEdge unifies cloud and edge under one Kubernetes control plane, so you use the same kubectl commands, Helm charts, and CI/CD pipelines for both. The offline autonomy feature ensures edge workloads keep running during network outages without manual intervention.

§03

How to use

  1. Install the cloud-side component on your Kubernetes cluster:
wget https://github.com/kubeedge/kubeedge/releases/download/v1.17.0/keadm-v1.17.0-linux-amd64.tar.gz
tar xzf keadm-v1.17.0-linux-amd64.tar.gz
cd keadm-v1.17.0-linux-amd64/keadm
./keadm init
  1. Join an edge node:
# On the edge device
./keadm join --cloudcore-ipport=<cloud-ip>:10000 --token=<token>
  1. Deploy workloads to edge nodes using standard Kubernetes node selectors or labels.
§04

Example

Deploying a container to edge nodes:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: edge-sensor-reader
spec:
  replicas: 1
  selector:
    matchLabels:
      app: sensor-reader
  template:
    metadata:
      labels:
        app: sensor-reader
    spec:
      nodeSelector:
        node-role.kubernetes.io/edge: ''
      containers:
        - name: reader
          image: my-registry/sensor-reader:v1
          resources:
            limits:
              memory: 128Mi
              cpu: 100m
§05

Related on TokRepo

§06

Common pitfalls

  • Edge nodes require at least 256MB RAM and 1 CPU core; very small devices (Arduino-class) are not supported
  • Network policies and service mesh features may not work identically on edge nodes due to the lightweight runtime
  • Upgrading KubeEdge components on disconnected edge nodes requires planning; stage updates when connectivity is available

Frequently Asked Questions

What is the difference between KubeEdge and standard Kubernetes?+

Standard Kubernetes assumes reliable network connectivity between nodes. KubeEdge adds an edge-aware architecture where edge nodes can operate autonomously when disconnected from the cloud control plane. It also reduces the resource footprint of the node agent.

Does KubeEdge support IoT device management?+

Yes. KubeEdge includes a device management framework that models IoT devices as Kubernetes custom resources. You can define device twins (digital representations) and communicate with devices using MQTT or other IoT protocols.

Can KubeEdge run on ARM devices?+

Yes. KubeEdge supports ARM64 and ARMv7 architectures, making it suitable for Raspberry Pi, NVIDIA Jetson, and similar edge hardware. Pre-built binaries are available for these architectures.

How does offline autonomy work?+

KubeEdge caches pod specifications and configuration on the edge node. When the network connection to the cloud is lost, the edge agent continues running existing workloads using the cached state. When connectivity returns, it syncs any changes.

Is KubeEdge production-ready?+

KubeEdge is a CNCF incubating project used in production by organizations in manufacturing, telecom, and transportation. It has been deployed in large-scale edge computing environments with thousands of nodes.

Citations (3)
Related on TokRepo
DevOps toolsMonitoring toolsFeatured workflows

Discussion

Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.

Related Assets

Conda — Cross-Platform Package and Environment Manager

Install, update, and manage packages and isolated environments for Python, R, C/C++, and hundreds of other languages from a single tool.

AI Open Source

Sphinx — Python Documentation Generator

Generate professional documentation from reStructuredText and Markdown with cross-references, API autodoc, and multiple output formats.

AI Open Source

Neutralinojs — Lightweight Cross-Platform Desktop Apps

Build desktop applications with HTML, CSS, and JavaScript using a tiny native runtime instead of bundling Chromium.

AI Open Source
◈Home🔍Search👤Me