Leaving no Leaf Device Behind: At the Frontier of Edge Computing
summary
Describes using Akri (CNCF Sandbox) to extend Kubernetes so workloads can be discovered and scheduled onto nodes that have real physical sensors. Demonstrates a live flow using k3s, Rancher, Elemental and Metal Cube to manage lightweight clusters: an Arduino is detected by Akri’s udev handler, a Deployment targets that device and streams readings to InfluxDB. Also reviews built-in handlers (udev, ONVIF, OPC), a Rust custom-handler template, GitOps-driven rollout, Raspberry Pi support and SUSE tech‑preview status.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:05Edge computing definition
- 3:15Edge types: near far tiny
- 5:01Edge vs IoT embedded
- 7:08Key edge challenges
- 9:36Kubernetes for edge
- 12:46Acri edge overview
- 15:14Acri device discovery
- 20:13Deploy Acri Helm GitOps
- 23:08Demo Arduino Acri InfluxDB
- 31:32Acri beyond edge
- 34:06Q&A
related talks
23mAkri: Making IoT Devices Accessible to Your Edge Kubernetes Clusters
39mKubernetes IoT Edge Working Group: Edge Device Onboa...
27mBridge Your IoT Leaf Devices to Local Clusters with Ease Using Akri...
36mEdge Computing in Kubernetes: Best Practices for the Future
10mOpening Remarks
35mBuilding Smart Farms with Edge Computing Systems Using Kubernetes
01h 16mIntro + Deep Dive: Kubernetes IoT Edge WG
39mIoT Application Running on KubeEdge + ARM Platform