Fabrikam inc has created a new operations team, and under its organization there is a brown field app called Drone Delivery. This application been running for a while in AKS (Kubernetes), and while they are huge fans of containers to build microservices and K8s, it has been discovered that it is not making use of any of the advance features like custom Service Mesh or Autoscaling among others.
The team has detected an opportunity to be more efficient at the devops level, and this is why they are now looking into a new fully managed Container App service to experiment with Fabrikam Drone Delivery. This will allow them to publish and run containarized microservices at scale, faster than before, reducing the complexity, saving resources by using scale to 0
built-in autoscaling capability, and without losing all the container advantages they love.
Azure Container Apps is a new cloud native serverless managed service that is just using AKS with KEDA behind the scenes to deploy and run containerized applications.
This repository guides you during the process of running an example application composed of microservices in Azure Container Apps. In this example scenario, the Fabrikam Drone Delivery app that was previously running in Azure Kubernetes Services will be run in a newly created Azure Container App environment. This Azure managed service is optimized for running applications that span many microservices. This example will make some containers internet-facing via an HTTPS ingress, and internally accessible thanks to its built-in DNS-based service discovery capability. Additionally, it will manage their secrets in a secure manner and authenticate against Azure KeyVault resources using User Managed Identities.
β Workflow Service is a message consumer app, so it needs to be deployed in single revision mode, otherwise an old versions could still process a message if it happens to be the one that retrieves it first. The rest of the microservices are configured with multiple revision mode.
For more information on how the Container Apps feature are being used in this Reference Implementation, please take a look below:
- HTTPS ingress, this allows to expose the Ingestion service to internet.
- Internal service discovery, Delivery, DroneScheduler and Package services must be internally reachable by Workflow service
- Use user-assigned identities when authenticating into Azure KeyVault from Delivery and DroneScheduler services
- Securely manage secrets for Package, Ingestion and Workflow services
- Run containers from any registry, the Fabrikam Drone Delivery uses ACR to publish its Docker images
- Use ARM templates to deploy my application, there is no need for another layer of indirection like Helm charts. All the Drone Delivery containers are part of the ARM templates
- Logs, see the container logs directly in Log Analytics without configuring any provider from code or Azure service.
-
An Azure subscription. You can open an account for free.
-
Azure CLI installed or you can perform this from Azure Cloud Shell by clicking below.
az login
-
Ensure you have latest version
az upgrade
Following the steps below will result in the creation of the following Azure resources that will be used throughout this Example Scenario.
Object | Purpose |
---|---|
An Azure Container App Environment | This is the managed Container App environment where Container Apps are deployed |
Five Azure Container Apps | These are the Azure resources that represents the five Fabrikam microservices in the Azure Container App environment |
An Azure Container Registry | This is the private container registry where all Fabrikam workload images are uploaded and later pulled from the different Azure Container Apps |
An Azure Log Analytics Workspace | This is where all the Container Apps logs are sent |
An Azure Application Insights instance | All services are sending trace information to a shared Azure Application Insights instance |
Two Azure Cosmos Db instances | Delivery and Package services have dependencies on Azure Cosmos DB |
An Azure Redis Cache instance | Delivery service uses Azure Redis cache to keep track of inflight deliveries |
An Azure Service Bus | Ingestion and Workflow services communicate using Azure Service Bus queues |
Five Azure User Managed Identities | These are going to give Read and List secrets permissions over Azure KeyVault to the microservices. |
Five Azure KeyVault instances | Secrets are saved into Azure KeyVault instances. |
-
Clone this repository
git clone --recurse-submodules https://github.com/mspnp/container-apps-fabrikam-dronedelivery.git
π‘ The steps shown here and elsewhere in the reference implementation use Bash shell commands. On Windows, you can install Windows Subsystem for Linux to run Bash by entering the following command in PowerShell or Windows Command Prompt and then restarting your machine:
wsl --install
-
Navigate to the container-apps-fabrikam-dronedelivery folder
cd ./container-apps-fabrikam-dronedelivery
-
Deploy the workload's prerequisites
az deployment sub create --name workload-stamp-prereqs --location eastus --template-file ./workload/workload-stamp-prereqs.json -p resourceGroupLocation=eastus
-
Get the workload User Assigned Identities
DELIVERY_PRINCIPAL_ID=$(az identity show -g rg-shipping-dronedelivery -n uid-delivery --query principalId -o tsv) && \ DRONESCHEDULER_PRINCIPAL_ID=$(az identity show -g rg-shipping-dronedelivery -n uid-dronescheduler --query principalId -o tsv) && \ WORKFLOW_PRINCIPAL_ID=$(az identity show -g rg-shipping-dronedelivery -n uid-workflow --query principalId -o tsv) && \ PACKAGE_ID_PRINCIPAL_ID=$(az identity show -g rg-shipping-dronedelivery -n uid-package --query principalId -o tsv) && \ INGESTION_ID_PRINCIPAL_ID=$(az identity show -g rg-shipping-dronedelivery -n uid-ingestion --query principalId -o tsv)
Warning As part of this initial migration, only Delivery and DroneScheduler services are making actual use of the Manage Identities to access their Azure KeyVault instances.
-
Deploy the workload Azure Container Registry and Azure resources associated to them
az deployment group create -f ./workload/workload-stamp.json -g rg-shipping-dronedelivery -p droneSchedulerPrincipalId=$DRONESCHEDULER_PRINCIPAL_ID \ -p workflowPrincipalId=$WORKFLOW_PRINCIPAL_ID \ -p deliveryPrincipalId=$DELIVERY_PRINCIPAL_ID \ -p ingestionPrincipalId=$INGESTION_ID_PRINCIPAL_ID \ -p packagePrincipalId=$PACKAGE_ID_PRINCIPAL_ID
-
Obtain the ACR server details
ACR_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.acrName.value -o tsv) ACR_SERVER=$(az acr show -n $ACR_NAME --query loginServer -o tsv) az acr update -n $ACR_NAME --admin-enabled true ACR_PASS=$(az acr credential show -n $ACR_NAME --query "passwords[0].value" -o tsv)
-
Build the microservice images
az acr build -r $ACR_NAME -t $ACR_SERVER/shipping/delivery:0.1.0 ./workload/src/shipping/delivery/. az acr build -r $ACR_NAME -t $ACR_SERVER/shipping/ingestion:0.1.0 ./workload/src/shipping/ingestion/. az acr build -r $ACR_NAME -t $ACR_SERVER/shipping/workflow:0.1.0 ./workload/src/shipping/workflow/. az acr build -r $ACR_NAME -f ./workload/src/shipping/dronescheduler/Dockerfile -t $ACR_SERVER/shipping/dronescheduler:0.1.0 ./workload/src/shipping/. az acr build -r $ACR_NAME -t $ACR_SERVER/shipping/package:0.1.0 ./workload/src/shipping/package/.
-
Get Application Insights instrumentation key
AI_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.appInsightsName.value -o tsv) AI_KEY=$(az resource show -g rg-shipping-dronedelivery -n $AI_NAME --resource-type "Microsoft.Insights/components" --query properties.InstrumentationKey -o tsv) AI_ID=$(az resource show -g rg-shipping-dronedelivery -n $AI_NAME --resource-type "Microsoft.Insights/components" --query properties.AppId -o tsv)
-
Get microservices details
# delivery DELIVERY_COSMOSDB_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.deliveryCosmosDbName.value -o tsv) DELIVERY_DATABASE_NAME="${DELIVERY_COSMOSDB_NAME}-db" DELIVERY_COLLECTION_NAME="${DELIVERY_COSMOSDB_NAME}-col" DELIVERY_COSMOSDB_ENDPOINT=$(az cosmosdb show -g rg-shipping-dronedelivery -n $DELIVERY_COSMOSDB_NAME --query documentEndpoint -o tsv) DELIVERY_REDIS_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.deliveryRedisName.value -o tsv) DELIVERY_REDIS_ENDPOINT=$(az redis show -g rg-shipping-dronedelivery -n $DELIVERY_REDIS_NAME --query hostName -o tsv) DELIVERY_KEYVAULT_URI=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.deliveryKeyVaultUri.value -o tsv) # drone scheduler DRONESCHEDULER_COSMOSDB_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.droneSchedulerCosmosDbName.value -o tsv) DRONESCHEDULER_COSMOSDB_ENDPOINT=$(az cosmosdb show -g rg-shipping-dronedelivery -n $DRONESCHEDULER_COSMOSDB_NAME --query documentEndpoint -o tsv) DRONESCHEDULER_KEYVAULT_URI=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.droneSchedulerKeyVaultUri.value -o tsv) # workflow WORKFLOW_NAMESPACE_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.ingestionQueueNamespace.value -o tsv) WORKFLOW_NAMESPACE_ENDPOINT=$(az servicebus namespace show -g rg-shipping-dronedelivery -n $WORKFLOW_NAMESPACE_NAME --query serviceBusEndpoint -o tsv) WORKFLOW_NAMESPACE_SAS_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.workflowServiceAccessKeyName.value -o tsv) WORKFLOW_NAMESPACE_SAS_KEY=$(az servicebus namespace authorization-rule keys list -g rg-shipping-dronedelivery --namespace-name $WORKFLOW_NAMESPACE_NAME -n $WORKFLOW_NAMESPACE_SAS_NAME --query primaryKey -o tsv) WORKFLOW_QUEUE_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.ingestionQueueName.value -o tsv) # package PACKAGE_MONGODB_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.packageMongoDbName.value -o tsv) PACKAGE_MONGODB_CONNNECTIONSTRING=$(az cosmosdb keys list --type connection-strings -g rg-shipping-dronedelivery --name $PACKAGE_MONGODB_NAME --query "connectionStrings[0].connectionString" -o tsv | sed 's/==/%3D%3D/g') # ingestion INGESTION_NAMESPACE_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.ingestionQueueNamespace.value -o tsv) INGESTION_NAMESPACE_SAS_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.ingestionServiceAccessKeyName.value -o tsv) INGESTION_NAMESPACE_SAS_KEY=$(az servicebus namespace authorization-rule keys list -g rg-shipping-dronedelivery --namespace-name $INGESTION_NAMESPACE_NAME -n $INGESTION_NAMESPACE_SAS_NAME --query primaryKey -o tsv) INGESTION_QUEUE_NAME=$(az deployment group show -g rg-shipping-dronedelivery -n workload-stamp --query properties.outputs.ingestionQueueName.value -o tsv)
-
Register the Azure Resource Manager provider for
Microsoft.App
az provider register --namespace Microsoft.App
-
Deploy the Container Apps ARM template
az deployment group create -f main.bicep -g rg-shipping-dronedelivery -p \ acrSever=$ACR_SERVER \ containerRegistryUser=$ACR_NAME \ containerRegistryPassword=$ACR_PASS \ applicationInsightsInstrumentationKey=$AI_KEY \ deliveryCosmosdbDatabaseName=$DELIVERY_DATABASE_NAME \ deliveryCosmosdbCollectionName=$DELIVERY_COLLECTION_NAME \ deliveryCosmosdbEndpoint=$DELIVERY_COSMOSDB_ENDPOINT \ deliveryRedisEndpoint=$DELIVERY_REDIS_ENDPOINT \ deliveryKeyVaultUri=$DELIVERY_KEYVAULT_URI \ droneSchedulerCosmosdbEndpoint=$DRONESCHEDULER_COSMOSDB_ENDPOINT \ droneSchedulerKeyVaultUri=$DRONESCHEDULER_KEYVAULT_URI \ wokflowNamespaceEndpoint=$WORKFLOW_NAMESPACE_ENDPOINT \ workflowNamespaceSASName=$WORKFLOW_NAMESPACE_SAS_NAME \ workflowNamespaceSASKey=$WORKFLOW_NAMESPACE_SAS_KEY \ workflowQueueName=$WORKFLOW_QUEUE_NAME \ packageMongodbConnectionString=$PACKAGE_MONGODB_CONNNECTIONSTRING \ ingestionNamespaceName=$INGESTION_NAMESPACE_NAME \ ingestionNamespaceSASName=$INGESTION_NAMESPACE_SAS_NAME \ ingestionNamespaceSASKey=$INGESTION_NAMESPACE_SAS_KEY \ ingestionQueueName=$INGESTION_QUEUE_NAME
π Please note that Azure Container Apps as well as this ARM API specification currently have limited
location
support.
Now that you have deployed in a Container Apps Environment, you can validate its functionality. This section will help you to validate the workload is exposed through a Container Apps External Ingress and responding to HTTP requests correctly.
-
Get the Ingestion FQDN
π The app team conducts a final acceptance test to ensure that traffic is flowing end-to-end as expected. To do so, an HTTP request is submitted against the ingestion external ingress.
INGESTION_FQDN=$(az deployment group show -g rg-shipping-dronedelivery -n main --query properties.outputs.ingestionFqdn.value -o tsv)
-
Send a request to https://dronedelivery.fabrikam.com.
π‘ Since the certificate used for TLS is self-signed, the request disables TLS validation using the '-k' option.
curl -X POST "https://${INGESTION_FQDN}/api/deliveryrequests" --header 'Content-Type: application/json' --header 'Accept: application/json' -d '{ "confirmationRequired": "None", "deadline": "", "dropOffLocation": "drop off", "expedited": true, "ownerId": "myowner", "packageInfo": { "packageId": "mypackage", "size": "Small", "tag": "mytag", "weight": 10 }, "pickupLocation": "mypickup", "pickupTime": "'$(date -u +%FT%TZ)'" }'
The response to the request printed in your terminal should look similar to the one shown below:
{"deliveryId":"5453d09a-a826-436f-8e7d-4ff706367b04","ownerId":"myowner","pickupLocation":"mypickup","pickupTime":"2021-02-14T20:00:00.000+0000","deadline":"","expedited":true,"confirmationRequired":"None","packageInfo":{"packageId":"mypackage","size":"Small","weight":10.0,"tag":"mytag"},"dropOffLocation":"drop off"}
-
Query Application Insights to ensure your request have been ingested by the underlaying services
β±οΈ It might take five minutes for the query results to be available.
az monitor app-insights query --app $AI_ID --analytics-query 'requests | summarize count_=sum(itemCount) by operation_Name | order by count_ desc | project strcat(operation_Name," (", count_, ")")' --query tables[0].rows[] -o table
The following output demonstrates the type of response to expect from the CLI command.
Result -------------------------------------------------- POST IngestionController/scheduleDeliveryAsync (1) PUT Deliveries/Put [id] (1) PUT /api/packages/mypackage (1) GET /api/packages/mypackage (1) PUT DroneDeliveries/Put [id] (1)
π Above result demonstrates that the http request initiated from the client has been ingested by
IngestionController/scheduleDeliveryAsync
to be later consumed by theWorkflow
background process to be sent toDeliveries/Put
,/api/packages/mypackage
andDroneDeliveries/Put
endpoints respectively. Them all are microservices running within Azure Container Apps.
If you need a restart a revision with Provision Status Failed
or for another reason you can use az cli:
az containerapp revision restart -g rg-shipping-dronedelivery --app <containerapp-name> -n <containerapp-revision-name>
-
Delete the Azure Container Registry resource group
az group delete -n rg-shipping-dronedelivery-acr -y
-
Delete the Azure Container Apps resource group
az group delete -n rg-shipping-dronedelivery -y
-
Delete the stored deployment data at subscription level
az deployment sub delete -n workload-stamp-prereqs
The team has been able to migrate and run Fabrikam Drone Delivery on top of Azure Container Apps. They are now laying out a new migration and modernization plan that will include:
Please see our contributor guide.
This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.