Using Kubernetes in IOT Edge Node Development
summary
Edge nodes run on small Intel NUCs with Ubuntu, each NUC using an external USB for local storage and ingesting vibration sensor streams that prove low data volume but enough to build pump-failure models. The prototype stacks a VM-hosted MongoDB + Spark + Zeppelin now moving to containers with Docker and a planned Kubernetes three-node cluster for high availability, CI/CD and rollbacks. Monitoring uses Monit feeding Slack, connectivity and on-site gateway issues drive the design constraints and choice to avoid cloud-managed services.
AI-written summary · about
chapters (18)
- 0:00Introduction and background
- 3:39Project origins and challenges
- 8:14Hardware selection and IoT setup
- 10:44Data collection and sensor use case
- 13:08Monitoring and cost constraints
- 14:52Upgrading hardware and sensor placement
- 16:14Machine learning and sensor networks
- 16:55Gateway reliability and connectivity issues
- 19:01Problems with virtual machines
- 20:43Why switch to containers
- 23:09Containers vs virtual machines
- 25:57Why Kubernetes for IoT edge
- 27:41Kubernetes benefits and automation
- 29:07IoT stack on Kubernetes
- 31:56Development workflow and GitOps
- 32:39Kubernetes IoT Edge working group
- 33:44Recap and key takeaways
- 34:26Challenge and next steps
related talks
36mKubernetes On the Edge With K3s For a Smart Metering Use Case
34mKubernetes IoT and Edge Working Group
01h 16mIntro + Deep Dive: Kubernetes IoT Edge WG
34mKubernetes IoT and Edge Working Group
39mIoT Application Running on KubeEdge + ARM Platform
34mManaging Edge Nodes With “KubeEdge”
25mKubernetes IoT Edge Working Group: Using Event Driven Architecture at Edge
34mKubeEdge: Kubernetes Native Edge Computing Framework