Overview
A Compute Instance provides a managed, on-demand development environment for data scientists and machine learning engineers. It comes pre‑installed with popular frameworks, SDKs, and tools, and integrates seamlessly with Azure Machine Learning Studio.
Prerequisites
- An active Azure subscription
- Azure Machine Learning workspace provisioned
- Permissions to create resources in the workspace
Step‑by‑Step Guide
1. Open Azure Machine Learning Studio
Navigate to ml.azure.com and select your workspace.
2. Create a Compute Instance
- In the left navigation pane, click Compute, then the Instances tab.
- Click New and fill in the required fields:
- Name: Choose a unique name.
- Region: Select the same region as your workspace.
- VM size: Pick a size based on your workload (e.g.,
Standard_DS3_v2).
- Optionally enable Auto‑shutdown to save costs.
- Click Create. Provisioning typically takes a few minutes.
3. Connect to the Instance
Once the instance status shows Running, click Start Jupyter or Start VS Code to launch your development environment directly in the browser.
Sample Azure CLI Script
Automate the creation of a Compute Instance with Azure CLI:
az ml compute create \
--name my‑compute‑instance \
--type ComputeInstance \
--size Standard_DS3_v2 \
--resource-group myResourceGroup \
--workspace-name myWorkspace
Next Steps
- Explore pre‑installed samples in AzureML examples repo.
- Configure Git integration for version control.
- Scale out training jobs using Compute Clusters.