Large-Scale Edge Management with Kubernetes: The Tech Behind Stable Operation of...
summary
PowerX runs ~200 EV chargers and builds a hybrid GKE-edge Kubernetes control plane using k0s to support ARMv7 and small memory footprints. Edge nodes join as lightweight workers and services deploy via Argo CD/GitOps as a daemonset for consistent deployment and self-healing. Metrics use two patterns: an on-edge OpenTelemetry collector that pushes to Prometheus when devices have ~1–4 GB, or a cloud-side scraper that pulls for 500 MB devices. System runs 3 years in production but costs ~700 MB/day network per device and adds troubleshooting complexity.
AI-written summary · about
related talks
36mEdge Computing in Kubernetes: Best Practices for the Future
06mEdge Kubernetes in Action: From Connected Vessels to Industrial Au...
34mKubeEdge: Kubernetes Native Edge Computing Framework
20mKubernetes at the Edge
31mIntro to KubeEdge: Kubernetes Native Edge Computing Framework
27mManaging Large-Scale Edge Cluster Over Unstable Network with KubeEdge
34mManaging Edge Nodes With “KubeEdge”
10mDevelop once, deploy anywhere: flexible infrastructure extends to the edge