Explore the versatility of KubeRay for managing Ray clusters on Kubernetes in this 33-minute conference talk. Learn how to deploy KubeRay and create your first cluster with a single command, then delve into advanced use cases beyond basic Kubectl and Helm deployments. Discover the extended API for end-user cluster management and client libraries that enable applications to scale Ray clusters on demand or programmatically spin up and manage new clusters. Get acquainted with the Python-Client KubeRay library, which simplifies cluster setup without the need for direct interaction with Kubernetes, Kubectl, or YAML files. By the end of the talk, gain a comprehensive understanding of Ray clustering on Kubernetes and the various mechanisms offered by the KubeRay project for cluster creation and management.
Overview
Syllabus
The Different Shades of using KubeRay with Kubernetes
Taught by
Anyscale