HOTSPOT You develop a Python application for image rendering. The application uses GPU resources to optimize rendering processes. You have the following requirements: • The application must be deployed to a Linux container. • The container must be stopped when the image rendering is complete.

Compute target: Azure Container Instances (ACI) is the most suitable choice. ACI is a serverless container-on-demand service ideal for workloads that run to completion, like a rendering job. It allows for the provisioning of single containers with GPU resources without the overhead of managing a full orchestration platform like Azure Kubernetes Service (AKS). The requirement to stop the container upon task completion aligns perfectly with ACI's task-based model.

Container termination: The Restart policy setting directly controls the lifecycle of the container after its main process exits. To ensure the container stops and does not restart after the rendering is finished, the restart policy should be set to OnFailure or Never. This ensures that once the application successfully completes its task (exits with code 0), the container instance stops as required.

Azure Container Instances (ACI) Documentation:

Microsoft Learn, "What is Azure Container Instances?". This document describes ACI as "a great solution for any scenario that can operate in isolated containers, including simple applications, task automation, and build jobs."

Microsoft Learn, "Run GPU-intensive workloads in Container Instances". This section confirms ACI's support for GPU resources, which is a key requirement. (Section: "GPU support")

Microsoft Learn, "Tutorial: Deploy a container application with a custom container image". This tutorial details the restartPolicy property for ACI container groups. It states: "The value of Never or OnFailure is best for task-based containers that run to completion." (Section: "Deploy the container")

Comparison of Azure Container Options:

Microsoft Learn, "Choose an Azure compute service for your application". This document compares services. For "Batch processing," it explicitly recommends ACI for "run-to-completion jobs." For AKS, it's recommended for "Orchestration" and "Long-running workloads," which is not the primary scenario described.