Azure Machine Learning Enterprise Terraform Example

This repo shows an example for rolling out a complete Azure Machine Learning enterprise enviroment via Terraform.

Deployed resources

This includes rollout of the following resources:

  • Azure Machine Learning Workspace with Private Link
  • Azure Storage Account with VNET binding (using Service Endpoints) and Private Link for Blob and File
  • Azure Key Vault with VNET binding (using Service Endpoints) and Private Link
  • Azure Container Registry
  • Azure Application Insights
  • Virtual Network
  • Jumphost (Windows) with Bastion for easy access to the VNET
  • Compute Cluster (in VNET)
  • Compute Instance (in VNET)
  • (Azure Kubernetes Service - disabled by default and still under development)

Instructions

Make sure you have the Azure CLI and the Azure Machine Learning CLI extension installed (az extension add -n azure-cli-ml).

  1. Copy terraform.tfvars.example to terraform.tfvars
  2. Update terraform.tfvars with your desired values
  3. Run Terraform 
     $ terraform init
 $ terraform plan
 $ terraform apply

If you want to deploy AKS, you need to have Azure Machine Learning CLI installed.

Known Limitations

  • Still need to update Default Datastore to use Managed Identity of the Studio UI data access

Important Notes

  • The user fileshare for the Compute Instances will be automatically provisioned upon first instance access