Kuberay 1.1: Distributed Computing Framework Enhancements

Introduction

Kuberay 1.1, released on November 28, 2024, advances distributed computing on Kubernetes with improved Ray cluster management, enhanced autoscaling, better performance, and expanded integration capabilities. This release makes Kuberay more powerful for running distributed computing workloads.

Ray Cluster Management

• Cluster creation improvements provide better Ray cluster creation and management.
• Node management enhancements enable better Ray node management.
• Configuration improvements enable better cluster configuration.
• Monitoring enhancements provide better visibility into cluster health.

Autoscaling Enhancements

1. Autoscaling improvements provide better automatic scaling of Ray clusters.
2. Resource optimization enables better resource utilization.
3. Scaling policies enable flexible scaling strategies.
4. Cost optimization reduces compute costs through efficient scaling.

Performance Optimizations

• Startup time reductions minimize Ray cluster startup time.
• Throughput improvements enable better handling of high-workload scenarios.
• Resource usage optimizations reduce CPU and memory consumption.
• Network efficiency improvements reduce network overhead.

Integration Capabilities

• Kubernetes integration improvements provide seamless integration with Kubernetes.
• Ray integration enables better integration with Ray framework.
• ML framework integration enables integration with ML frameworks.
• API improvements enable better programmatic access to Kuberay capabilities.

Getting Started

# Install Kuberay operator
kubectl apply -k "github.com/ray-project/kuberay/ray-operator/config/default"

Create a RayCluster:

apiVersion: ray.io/v1
kind: RayCluster
metadata:
  name: raycluster
spec:
  rayVersion: '2.8.0'
  enableInTreeAutoscaling: true
  autoscalerOptions:
    upscalingMode: Default
    idleTimeoutSeconds: 60
  headGroupSpec:
    serviceType: ClusterIP
    rayStartParams:
      dashboard-host: '0.0.0.0'
    template:
      spec:
        containers:
        - name: ray-head
          image: rayproject/ray:2.8.0
          ports:
          - containerPort: 6379
          - containerPort: 8265
          - containerPort: 10001
          resources:
            limits:
              cpu: "1"
              memory: "1Gi"
            requests:
              cpu: "1"
              memory: "1Gi"
  workerGroupSpecs:
  - replicas: 1
    minReplicas: 1
    maxReplicas: 5
    groupName: small-group
    rayStartParams: {}
    template:
      spec:
        containers:
        - name: ray-worker
          image: rayproject/ray:2.8.0
          resources:
            limits:
              cpu: "1"
              memory: "1Gi"
            requests:
              cpu: "1"
              memory: "1Gi"

Summary

Kuberay 1.1 continues to evolve as a leading distributed computing solution, providing teams with powerful capabilities for running distributed computing workloads on Kubernetes.