Bridging the Local Kubernetes Gap for...
summary
vCluster in Docker (wind) provides a local Kubernetes that actually mirrors platform needs: it gives inbuilt load balancer support, a platform UI and a demoed workflow that lets a laptop control multi-node clusters from a single config. Developers can attach real GPU nodes over VPN (any Nvidia machine or cloud instance joins via a vcluster token), use a pull‑through cache, and sleep and wake clusters for fast feedback. The system is open source and the UI demo shows it working end‑to‑end.
AI-written summary · about
related talks
01h 22mWorkshop: Optimizing Kubernetes Cluster Scaling for Adv...
22mBeyond Single-Cluster Limits: Scaling GPU Workloads Across Kubernetes With…
32mGPU Machine Learning From Laptop to Cloud
06mKubernetes Love Machine Learning, Even on Private Cloud
11mGPU‑Accelerated Workloads on KubeVirt: Scaling ML/AI in Kuberne...
30mRunning AI Workloads in Containers and Kubernetes
36mMulti-Cloud Machine Learning Data and Workflow with Kubernetes
38mBeyond Kubernetes: Adapting To Spe...