AKS Anyscale Setup

Automated multi-region Azure Kubernetes Service (AKS) cluster setup configured for Ray workloads with GPU support.

Overview

This repository provides infrastructure-as-code automation to deploy production-ready AKS clusters across multiple Azure regions. Each cluster is configured with:

• Azure CNI with Overlay networking
• OIDC and Workload Identity for secure authentication
• Three node pools: system, CPU (on-demand), and GPU (spot instances with A100 GPUs)
• NGINX Ingress Controller with LoadBalancer
• NVIDIA Device Plugin for GPU workload scheduling
• Anyscale Operator for AI/ML workload orchestration

Prerequisites

• Azure CLI (az) installed and authenticated
• Anyscale CLI installed with credentials at ~/.anyscale/credentials.json
• Helm 3.x
• jq for JSON parsing

Quick Start

1. Clone the repository:

git clone <repository-url>
cd aks-anyscale

2. (Optional) Configure environment variables:

export ANYSCALE_CLOUD_NAME="my-cloud-name"
export REGIONS="eastus westus"
export PRIMARY_REGION="eastus"
export SUBSCRIPTION="your-subscription-id"
export RESOURCE_GROUP="my-rg"

3. Run the setup script:

./setup.sh

The script will provision all infrastructure including:

• Resource group and storage account
• Virtual networks with NAT gateways
• AKS clusters with three node pools
• Kubernetes add-ons (NGINX, NVIDIA Device Plugin)
• Anyscale Operator installation and cloud registration

Configuration

Default configuration can be overridden via environment variables. See setup.sh for all available options.

Helm Values

Configuration files for Kubernetes components:

• values_nginx.yaml: NGINX Ingress Controller settings
• values_nvidia.yaml: NVIDIA Device Plugin settings
• values_anyscale.yaml: Anyscale Operator custom instance types
• cloud_resource.yaml: Template for secondary region registration

Reference

• Anyscale Documentation
• AKS Documentation