Platform Orchestration: Anyscale + Ray on Kubernetes
summary
Anyscale on Azure integrates into AKS running in the customer's subscription with ARM-native provisioning, Entra ID workload identity, and data that never leaves the VNet. KubeRay Connect installs as a lightweight Helm chart, uses a mutating webhook to inject sidecars and an OpenTelemetry collector, splits control from telemetry, and in the demo the connector shows as live and running. Anyscale Scheduler unifies multi-cluster multi-cloud scheduling, retires fragile machine pools, and provides priority-aware queuing and on-demand capacity search.
AI-written summary · about
chapters (5)
- 0:00Introduction and industry trends
- 3:56Anyscale on Azure integration
- 7:10KubeRay Connect announcement
- 13:59Anyscale scheduler improvements
- 16:49Q&A and closing
related talks
31mManaging Large-Scale Ray Deployments: Cloud, On-Prem, and Kubernetes
01h 01mDevOps with Azure Kubernetes Service
24mSkyRay: Seamlessly Extending KubeRay to Multi-Cluster Multi-Cloud Operation
32mMastering Cloud Kubernetes: Navigating Azure's Kubernetes Service Maze
29mOrchestrating 10,000+ Kubernetes deployments with 1 pipeline
26mIntroduction to Distributed ML Workloads with Ray on Kubernetes
35mKubernetes journey
49mAZURE CONTAINERS AND KUBERNETES PLATFORM
