Deploy an AKS cluster with Confidential Containers and a default policy

Article
02/27/2024

In this article, you use the Azure CLI to deploy an Azure Kubernetes Service (AKS) cluster and configure Confidential Containers (preview) with a default security policy. You then deploy an application as a Confidential container. To learn more, read the overview of AKS Confidential Containers.

In general, getting started with AKS Confidential Containers involves the following steps.

Deploy or upgrade an AKS cluster using the Azure CLI
Add an annotation to your pod YAML manifest to mark the pod as being run as a confidential container
Add a security policy to your pod YAML manifest
Enable enforcement of the security policy
Deploy your application in confidential computing

Prerequisites

The Azure CLI version 2.44.1 or later. Run az --version to find the version, and run az upgrade to upgrade the version. If you need to install or upgrade, see Install Azure CLI.
The aks-preview Azure CLI extension version 0.5.169 or later.
The confcom Confidential Container Azure CLI extension 0.3.3 or later. confcom is required to generate a security policy.
Register the Preview feature in your Azure subscription.
AKS supports Confidential Containers (preview) on version 1.25.0 and higher.
A workload identity and a federated identity credential. The workload identity credential enables Kubernetes applications access to Azure resources securely with a Microsoft Entra ID based on annotated service accounts. If you aren't familiar with Microsoft Entra Workload ID, see the Microsoft Entra Workload ID overview and review how Workload Identity works with AKS.
The identity you're using to create your cluster has the appropriate minimum permissions. For more information about access and identity for AKS, see Access and identity options for Azure Kubernetes Service (AKS).
To manage a Kubernetes cluster, use the Kubernetes command-line client kubectl. You can install kubectl locally using the az aks install-cli command.
Confidential containers on AKS provide a sidecar open source container for attestation and secure key release. The sidecar integrates with a Key Management Service (KMS), like Azure Key Vault, for releasing a key to the container group after validation is completed.

Install the aks-preview Azure CLI extension

Important

AKS preview features are available on a self-service, opt-in basis. Previews are provided "as is" and "as available," and they're excluded from the service-level agreements and limited warranty. AKS previews are partially covered by customer support on a best-effort basis. As such, these features aren't meant for production use. For more information, see the following support articles:

To install the aks-preview extension, run the following command:

az extension add --name aks-preview

Run the following command to update to the latest version of the extension released:

az extension update --name aks-preview

Install the confcom Azure CLI extension

To install the confcom extension, run the following command:

az extension add --name confcom

Run the following command to update to the latest version of the extension released:

az extension update --name confcom

Register the KataCcIsolationPreview feature flag

Register the KataCcIsolationPreview feature flag by using the az feature register command, as shown in the following example:

az feature register --namespace "Microsoft.ContainerService" --name "KataCcIsolationPreview"

It takes a few minutes for the status to show Registered. Verify the registration status by using the az feature show command:

az feature show --namespace "Microsoft.ContainerService" --name "KataCcIsolationPreview"

When the status reflects Registered, refresh the registration of the Microsoft.ContainerService resource provider by using the az provider register command:

az provider register --namespace "Microsoft.ContainerService"

Deploy a new cluster

Create an AKS cluster using the az aks create command and specifying the following parameters:
- --os-sku: AzureLinux. Only the Azure Linux os-sku supports this feature in this preview release.
- --node-vm-size: Any Azure VM size that is a generation 2 VM and supports nested virtualization works. For example, [Standard_DC8as_cc_v5][DC8as-series] VMs.
- --enable-workload-identity: Enables creating a Microsoft Entra Workload ID enabling pods to use a Kubernetes identity.
- --enable-oidc-issuer: Enables OpenID Connect (OIDC) Issuer. It allows a Microsoft Entra ID or other cloud provider identity and access management platform the ability to discover the API server's public signing keys.
The following example updates the cluster named myAKSCluster and creates a single system node pool in the myResourceGroup:
```
az aks create --resource-group myResourceGroup --name myAKSCluster --kubernetes-version <1.25.0 and above> --os-sku AzureLinux --node-vm-size Standard_DC4as_cc_v5 --node-count 1 --enable-oidc-issuer --enable-workload-identity --generate-ssh-keys
```
After a few minutes, the command completes and returns JSON-formatted information about the cluster. The cluster created in the previous step has a single node pool. In the next step, we add a second node pool to the cluster.
When the cluster is ready, get the cluster credentials using the az aks get-credentials command.
```
az aks get-credentials --resource-group myResourceGroup --name myAKSCluster
```
Add a user node pool to myAKSCluster with two nodes in nodepool2 in the myResourceGroup using the az aks nodepool add command. Specify the following parameters:
- --workload-runtime: Specify KataCcIsolation to enable the Confidential Containers feature on the node pool. With this parameter, these other parameters shall satisfy the following requirements. Otherwise, the command fails and reports an issue with the corresponding parameter(s).
- --os-sku: AzureLinux. Only the Azure Linux os-sku supports this feature in this preview release.
- --node-vm-size: Any Azure VM size that is a generation 2 VM and supports nested virtualization works. For example, [Standard_DC8as_cc_v5][DC8as-series] VMs.
```
az aks nodepool add --resource-group myResourceGroup --name nodepool2 --cluster-name myAKSCluster --node-count 2 --os-sku AzureLinux --node-vm-size Standard_DC4as_cc_v5 --workload-runtime KataCcIsolation
```

After a few minutes, the command completes and returns JSON-formatted information about the cluster.

Deploy to an existing cluster

To use this feature with an existing AKS cluster, the following requirements must be met:

Follow the steps to register the KataCcIsolationPreview feature flag.
Verify the cluster is running Kubernetes version 1.25.0 and higher.
Enable workload identity on the cluster if it isn't already.

Use the following command to enable Confidential Containers (preview) by creating a node pool to host it.

Add a node pool to your AKS cluster using the az aks nodepool add command. Specify the following parameters:
- --resource-group: Enter the name of an existing resource group to create the AKS cluster in.
- --cluster-name: Enter a unique name for the AKS cluster, such as myAKSCluster.
- --name: Enter a unique name for your clusters node pool, such as nodepool2.
- --workload-runtime: Specify KataCcIsolation to enable the feature on the node pool. Along with the --workload-runtime parameter, these other parameters shall satisfy the following requirements. Otherwise, the command fails and reports an issue with the corresponding parameter(s).
- --os-sku: AzureLinux. Only the Azure Linux os-sku supports this feature in this preview release.
- --node-vm-size: Any Azure VM size that is a generation 2 VM and supports nested virtualization works.
The following example adds a user node pool to myAKSCluster with two nodes in nodepool2 in the myResourceGroup:
```
az aks nodepool add --resource-group myResourceGroup --name nodepool2 –-cluster-name myAKSCluster --node-count 2 --os-sku AzureLinux --node-vm-size Standard_DC4as_cc_v5 --workload-runtime KataCcIsolation
```
After a few minutes, the command completes and returns JSON-formatted information about the cluster.
Run the az aks update command to enable Confidential Containers (preview) on the cluster.
```
az aks update --name myAKSCluster --resource-group myResourceGroup
```
After a few minutes, the command completes and returns JSON-formatted information about the cluster.
When the cluster is ready, get the cluster credentials using the az aks get-credentials command.
```
az aks get-credentials --resource-group myResourceGroup --name myAKSCluster
```

Configure container

Before you configure access to the Azure Key Vault and secret, and deploy an application as a Confidential container, you need to complete the configuration of the workload identity.

To configure the workload identity, perform the following steps described in the Deploy and configure workload identity article:

Retrieve the OIDC Issuer URL
Create a managed identity
Create Kubernetes service account
Establish federated identity credential

Important

You need to set the environment variables from the section Export environmental variables in the Deploy and configure workload identity article to continue completing this tutorial. Remember to set the variable SERVICE_ACCOUNT_NAMESPACE to kafka, and execute the command kubectl create namespace kafka before configuring workload identity.

Deploy a trusted application with kata-cc and attestation container

The following steps configure end-to-end encryption for Kafka messages using encryption keys managed by Azure Managed Hardware Security Modules (mHSM). The key is only released when the Kafka consumer runs within a Confidential Container with an Azure attestation secret provisioning container injected in to the pod.

This configuration is based on the following four components:

Kafka Cluster: A simple Kafka cluster deployed in the Kafka namespace on the cluster.
Kafka Producer: A Kafka producer running as a vanilla Kubernetes pod that sends encrypted user-configured messages using a public key to a Kafka topic.
Kafka Consumer: A Kafka consumer pod running with the kata-cc runtime, equipped with a secure key release container to retrieve the private key for decrypting Kafka messages and render the messages to web UI.

For this preview release, we recommend for test and evaluation purposes to either create or use an existing Azure Key Vault Premium tier resource to support storing keys in a hardware security module (HSM). We don't recommend using your production key vault. If you don't have an Azure Key Vault, see Create a key vault using the Azure CLI.

Grant the managed identity you created earlier, and your account, access to the key vault. Assign both identities the Key Vault Crypto Officer and Key Vault Crypto User Azure RBAC roles.
Note
- The managed identity is the value you assign to the USER_ASSIGNED_IDENTITY_NAME variable.
- To add role assignments, you must have Microsoft.Authorization/roleAssignments/write and Microsoft.Authorization/roleAssignments/delete permissions, such as Key Vault Data Access Administrator, User Access Administrator, or Owner.
Run the following command to set the scope:
```
AKV_SCOPE=$(az keyvault show --name <AZURE_AKV_RESOURCE_NAME> --query id --output tsv)
```
Run the following command to assign the Key Vault Crypto Officer role.
```
az role assignment create --role "Key Vault Crypto Officer" --assignee "${USER_ASSIGNED_IDENTITY_NAME}" --scope $AKV_SCOPE
```
Run the following command to assign the Key Vault Crypto User role.
```
az role assignment create --role "Key Vault Crypto User" --assignee "${USER_ASSIGNED_IDENTITY_NAME}" --scope $AKV_SCOPE
```
Install the Kafka cluster in the kafka namespace by running the following command:
```
kubectl create -f 'https://strimzi.io/install/latest?namespace=kafka' -n kafka
```

Run the following command to apply the kafka cluster CR file.

kubectl apply -f https://strimzi.io/examples/latest/kafka/kafka-persistent-single.yaml -n kafka

Prepare the RSA Encryption/Decryption key using the bash script for the workload from GitHub. Save the file as setup-key.sh.
Set the MAA_ENDPOINT environment variable with the FQDN of Attest URI by running the following command.
```
export MAA_ENDPOINT="$(az attestation show --name "myattestationprovider" --resource-group "MyResourceGroup" --query 'attestUri' -o tsv | cut -c 9-)"
```
Check if the FQDN of Attest URI is in correct format (the MAA_ENDPOINT should not include the prefix "https://"):
```
echo $MAA_ENDPOINT
```
Note

To set up Azure Attestation, see [Quickstart: Set up Azure Attestation with Azure CLI][attestation-quickstart-azure-cli].

Copy the following YAML manifest and save it as consumer.yaml.

apiVersion: v1
kind: Pod
metadata:
  name: kafka-golang-consumer
  namespace: kafka
  labels:
    azure.workload.identity/use: "true"
    app.kubernetes.io/name: kafka-golang-consumer
spec:
  serviceAccountName: workload-identity-sa
  runtimeClassName: kata-cc-isolation
  containers:
    - image: "mcr.azk8s.cn/aci/skr:2.7"
      imagePullPolicy: Always
      name: skr
      env:
        - name: SkrSideCarArgs
          value: ewogICAgImNlcnRjYWNoZSI6IHsKCQkiZW5kcG9pbnRfdHlwZSI6ICJMb2NhbFRISU0iLAoJCSJlbmRwb2ludCI6ICIxNjkuMjU0LjE2OS4yNTQvbWV0YWRhdGEvVEhJTS9hbWQvY2VydGlmaWNhdGlvbiIKCX0gIAp9
      command:
        - /bin/skr
      volumeMounts:
        - mountPath: /opt/confidential-containers/share/kata-containers/reference-info-base64
          name: endor-loc
    - image: "mcr.azk8s.cn/acc/samples/kafka/consumer:1.0"
      imagePullPolicy: Always
      name: kafka-golang-consumer
      env:
        - name: SkrClientKID
          value: kafka-encryption-demo
        - name: SkrClientMAAEndpoint
          value: sharedeus2.eus2.test.attest.chinacloudapi.cn
        - name: SkrClientAKVEndpoint
          value: "myKeyVault.vault.azure.cn"
        - name: TOPIC
          value: kafka-demo-topic
      command:
        - /consume
      ports:
        - containerPort: 3333
          name: kafka-consumer
      resources:
        limits:
          memory: 1Gi
          cpu: 200m
  volumes:
    - name: endor-loc
      hostPath:
        path: /opt/confidential-containers/share/kata-containers/reference-info-base64
---
apiVersion: v1
kind: Service
metadata:
  name: consumer
  namespace: kafka
spec:
  type: LoadBalancer
  selector:
    app.kubernetes.io/name: kafka-golang-consumer
  ports:
    - protocol: TCP
      port: 80
      targetPort: kafka-consumer

Note

Update the value for the pod environment variable SkrClientAKVEndpoint to match the URL of your Azure Key Vault, excluding the protocol value https://. The current value placeholder value is myKeyVault.vault.azure.cn. Update the value for the pod environment variable SkrClientMAAEndpoint with the value of MAA_ENDPOINT. You can find the value of MAA_ENDPOINT by running the command echo $MAA_ENDPOINT or the command az attestation show --name "myattestationprovider" --resource-group "MyResourceGroup" --query 'attestUri' -o tsv | cut -c 9-.

Generate the security policy for the Kafka consumer YAML manifest and obtain the hash of the security policy stored in the WORKLOAD_MEASUREMENT variable by running the following command:
```
export WORKLOAD_MEASUREMENT=$(az confcom katapolicygen -y consumer.yaml --print-policy | base64 -d | sha256sum | cut -d' ' -f1)
```
To generate an RSA asymmetric key pair (public and private keys), run the setup-key.sh script using the following command. The <Azure Key Vault URL> value should be <your-unique-keyvault-name>.vault.azure.cn
```
export MANAGED_IDENTITY=${USER_ASSIGNED_CLIENT_ID}
bash setup-key.sh "kafka-encryption-demo" <Azure Key Vault URL>
```
Note
- The envionment variable MANAGED_IDENTITY is required by the bash script setup-key.sh.
- The public key will be saved as kafka-encryption-demo-pub.pem after executing the bash script.
Important

If you receive the error ForbiddenByRbac,you might need to wait up to 24 hours as the backend services for managed identities maintain a cache per resource URI for up to 24 hours. See also: Troubleshoot Azure RBAC.

To verify the keys have been successfully uploaded to the key vault, run the following commands:

az account set --subscription <Subscription ID>
az keyvault key list --vault-name <KeyVault Name> -o table

Copy the following YAML manifest and save it as producer.yaml.

apiVersion: v1
kind: Pod
metadata:
  name: kafka-producer
  namespace: kafka
spec:
  containers:
    - image: "mcr.azk8s.cn/acc/samples/kafka/producer:1.0"
      name: kafka-producer
      command:
        - /produce
      env:
        - name: TOPIC
          value: kafka-demo-topic
        - name: MSG
          value: "Azure Confidential Computing"
        - name: PUBKEY
          value: |-
            -----BEGIN PUBLIC KEY-----
            MIIBojAN***AE=
            -----END PUBLIC KEY-----
      resources:
        limits:
          memory: 1Gi
          cpu: 200m

Note

Update the value which begin with -----BEGIN PUBLIC KEY----- and ends with -----END PUBLIC KEY----- strings with the content from kafka-encryption-demo-pub.pem which was created in the previous step.

Deploy the consumer and producer YAML manifests using the files you saved earlier.
```
kubectl apply -f consumer.yaml
```
```
kubectl apply -f producer.yaml
```
Get the IP address of the web service using the following command:
```
kubectl get svc consumer -n kafka
```
Copy and paste the external IP address of the consumer service into your browser and observe the decrypted message.

The following example resembles the output of the command:
```
Welcome to Confidential Containers on AKS!
Encrypted Kafka Message:
Msg 1: Azure Confidential Computing
```
You should also attempt to run the consumer as a regular Kubernetes pod by removing the skr container and kata-cc runtime class spec. Since you aren't running the consumer with kata-cc runtime class, you no longer need the policy.
Remove the entire policy and observe the messages again in the browser after redeploying the workload. Messages appear as base64-encoded ciphertext because the private encryption key can't be retrieved. The key can't be retrieved because the consumer is no longer running in a confidential environment, and the skr container is missing, preventing decryption of messages.

Cleanup

When you're finished evaluating this feature, to avoid Azure charges, clean up your unnecessary resources. If you deployed a new cluster as part of your evaluation or testing, you can delete the cluster using the az aks delete command.

az aks delete --resource-group myResourceGroup --name myAKSCluster

If you enabled Confidential Containers (preview) on an existing cluster, you can remove the pod(s) using the kubectl delete pod command.

kubectl delete pod pod-name

Next steps

Learn more about Azure Dedicated hosts for nodes with your AKS cluster to use hardware isolation and control over Azure platform maintenance events.