Deploying the AKS Store Demo app to Azure using Azure Developer CLI

Using the Azure Developer CLI, you can deploy this solution to Azure in minutes. By default it ships prebuilt container images and RabbitMQ/MongoDB; you can also opt into Azure Service Bus and Azure Cosmos DB, and even build app images from source.

Prerequisites

Opening the AKS Store Demo repo in GitHub Codespaces is preferred; however, if you want to run the app locally, you will need the following tools:

• Azure CLI
• Azure Developer CLI version 1.15.0 or later
• Visual Studio Code
• Docker Desktop
• kubectl
• kubelogin
• Helm
• kustomize
• Git
• Terraform
• Bash shell

Get started

To get started, authenticate to Azure using the Azure Developer CLI and Azure CLI.

# authenticate to Azure Developer CLI
azd auth login

# enable Helm support
azd config set alpha.aks.helm on

# enable Kustomize support (used when building from source)
azd config set alpha.aks.kustomize on

# authenticate to Azure CLI
az login

az provider register --namespace Microsoft.AlertsManagement
az provider register --namespace Microsoft.CognitiveServices
az provider register --namespace Microsoft.ContainerService
az provider register --namespace Microsoft.Dashboard
az provider register --namespace Microsoft.DocumentDB
az provider register --namespace Microsoft.KeyVault
az provider register --namespace Microsoft.OperationalInsights
az provider register --namespace Microsoft.ServiceBus

When selecting an Azure region, choose one that supports all services used here: Azure OpenAI, AKS, Key Vault, Service Bus, Cosmos DB, Log Analytics, Azure Monitor (managed Prometheus), and Managed Grafana.

Availability zone support

For increased resiliency you may want to deploy into a region that supports Availability Zones. Availability zone mappings are assigned per subscription, so the set of zones available can vary between subscriptions and regions. You can use the following command to list regions that support all three availability zones (1, 2, and 3):

az account list-locations \
  --query "sort_by([? (availabilityZoneMappings != null && contains(availabilityZoneMappings[].logicalZone, '1') && contains(availabilityZoneMappings[].logicalZone, '2') && contains(availabilityZoneMappings[].logicalZone, '3')) || (metadata.availabilityZoneMappings != null && contains(metadata.availabilityZoneMappings[].logicalZone, '1') && contains(metadata.availabilityZoneMappings[].logicalZone, '2') && contains(metadata.availabilityZoneMappings[].logicalZone, '3')) ], &name)[].{Region:name}" \
  --output table

See the Azure documentation on availability zones for details and service-specific guidance.

If you are deploying an Azure OpenAI account, you will need to ensure you have enough tokens per minute quota for the gpt-5-mini model. You can check your quota by running the following command:

REGION=swedencentral

az cognitiveservices usage list \
  --location $REGION \
  --query "[].{name: name.value, currentValue:currentValue, limit: limit}" \
  -o table

Deployment settings

The infrastructure-as-code templates in this repo use variables to define the deployment settings. You can set these variables using the Azure Developer CLI and the templates will evaluate them to provision the resources.

The following environment variables control what gets deployed:

These environment variables listed above can be set with commands like this:

# set the main deployment location
azd env set AZURE_LOCATION swedencentral

# set the SKU of the virtual machine scale set nodes in the AKS cluster
azd env set AKS_NODE_POOL_VM_SIZE Standard_D2s_v6

# deploys azure container registry and imports containers from github container registry
azd env set DEPLOY_AZURE_CONTAINER_REGISTRY true

# deploys Azure OpenAI
azd env set DEPLOY_AZURE_OPENAI true

# Azure OpenAI region
azd env set AZURE_OPENAI_LOCATION swedencentral

# deploys azure service bus
azd env set DEPLOY_AZURE_SERVICE_BUS true

# deploys azure cosmos db with the sql api
azd env set DEPLOY_AZURE_COSMOSDB true

# set Cosmos DB account kind (GlobalDocumentDB for SQL API, MongoDB for MongoDB API)
azd env set AZURE_COSMOSDB_ACCOUNT_KIND GlobalDocumentDB

# deploys aks observability tools
azd env set DEPLOY_OBSERVABILITY_TOOLS true

# set custom source registry (optional)
azd env set SOURCE_REGISTRY ghcr.io/azure-samples

Deploy the app

Provision and deploy the app with a single command.

azd up

When you run the azd up command for the first time, you will be asked for a bit of information:

• Environment name: This is the name of the environment that will be created so that Azure Developer CLI can keep track of the resources that are created.
• Azure subscription: You will be asked to select the Azure subscription that you want to use. If you only have one subscription, it will be selected by default.
• Azure location: You will be asked to select the Azure location where the resources will be created. You can select the location that is closest to you but you must ensure that the location supports all the resources that will be created. If you are unsure of which region to use, select "East US 2".

After you provide the information, azd up registers providers/features and installs required Azure CLI extensions. It then runs Terraform to provision Azure resources and deploys the app to AKS using a Helm chart. Workload identity is configured automatically for services that talk to Azure resources.

This will take a few minutes to complete.

Build from source (optional)

For a full source-to-ACR build and Kustomize-based deploy:

1. Swap the azd config to the build-from-source variant

mv azure.yaml azure.yaml.bak
mv azure-build-from-source.yaml azure.yaml

1. Ensure ACR is enabled (images will be built and pushed there)

azd env set DEPLOY_AZURE_CONTAINER_REGISTRY true

This flow builds Docker images for each service and deploys using Kustomize overlays.

Validate the deployment

Once the deployment completes, azd prints outputs. You can get service URLs directly:

azd env get-value SERVICE_STORE_FRONT_ENDPOINT_URL
azd env get-value SERVICE_STORE_ADMIN_ENDPOINT_URL

You can also browse the resource group (AZURE_RESOURCE_GROUP) in the Azure Portal. In the AKS resource, check Workloads and Services/Ingresses in the pets namespace. store-front and store-admin are exposed via LoadBalancers with public IPs.

If you deployed an Azure Service Bus, navigate to the resource and use Azure Service Bus explorer to check for order messages.

If you deployed an Azure CosmosDB, navigate to the resource and use the database explorer to check for order records.

Clean up

When you are done testing the deployment, you can clean up the resources using the azd down command.

azd down --force --purge