Configure Azure CNI Powered by Cilium in Azure Kubernetes Service (AKS)

Azure CNI Powered by Cilium combines the robust control plane of Azure CNI with the data plane of Cilium to provide high-performance networking and security.

By making use of eBPF programs loaded into the Linux kernel and a more efficient API object structure, Azure CNI Powered by Cilium provides the following benefits:

• Functionality equivalent to existing Azure CNI and Azure CNI Overlay plugins

• Improved Service routing

• More efficient network policy enforcement

• Better observability of cluster traffic

• Support for larger clusters (more nodes, pods, and services)

IP Address Management (IPAM) with Azure CNI Powered by Cilium

Azure CNI Powered by Cilium can be deployed using two different methods for assigning pod IPs:

• Assign IP addresses from an overlay network (similar to Azure CNI Overlay mode)

• Assign IP addresses from a virtual network (similar to existing Azure CNI with Dynamic Pod IP Assignment)

If you aren't sure which option to select, read "Choosing a network model to use".

Network Policy Enforcement

Cilium enforces network policies to allow or deny traffic between pods. With Cilium, you don't need to install a separate network policy engine such as Azure Network Policy Manager or Calico.

Limitations

Azure CNI powered by Cilium currently has the following limitations:

• Available only for Linux and not for Windows.

• Cilium L7 policy enforcement is disabled.

• Network policies cannot use ipBlock to allow access to node or pod IPs. See frequently asked questions for details and recommended workaround.

• Multiple Kubernetes services can't use the same host port with different protocols (for example, TCP or UDP) (Cilium issue #14287).

• Network policies may be enforced on reply packets when a pod connects to itself via service cluster IP (Cilium issue #19406).

• Network policies are not applied to pods using host networking (spec.hostNetwork: true) because these pods use the host identity instead of having individual identities.

Prerequisites

• Azure CLI version 2.48.1 or later. Run az --version to see the currently installed version. If you need to install or upgrade, see Install Azure CLI.

• If using ARM templates or the REST API, the AKS API version must be 2022-09-02-preview or later.

Create a new AKS Cluster with Azure CNI Powered by Cilium

Option 1: Assign IP addresses from an overlay network

Use the following commands to create a cluster with an overlay network and Cilium. Replace the values for <clusterName>, <resourceGroupName>, and <location>:

az aks create \
    --name <clusterName> \
    --resource-group <resourceGroupName> \
    --location <location> \
    --network-plugin azure \
    --network-plugin-mode overlay \
    --pod-cidr 192.168.0.0/16 \
    --network-dataplane cilium \
    --generate-ssh-keys

Option 2: Assign IP addresses from a virtual network

Run the following commands to create a resource group and virtual network with a subnet for nodes and a subnet for pods.

# Create the resource group
az group create --name <resourceGroupName> --location <location>

# Create a virtual network with a subnet for nodes and a subnet for pods
az network vnet create --resource-group <resourceGroupName> --location <location> --name <vnetName> --address-prefixes <address prefix, example: 10.0.0.0/8> -o none
az network vnet subnet create --resource-group <resourceGroupName> --vnet-name <vnetName> --name nodesubnet --address-prefixes <address prefix, example: 10.240.0.0/16> -o none
az network vnet subnet create --resource-group <resourceGroupName> --vnet-name <vnetName> --name podsubnet --address-prefixes <address prefix, example: 10.241.0.0/16> -o none

Create the cluster using --network-dataplane cilium:

az aks create \
    --name <clusterName> \
    --resource-group <resourceGroupName> \
    --location <location> \
    --max-pods 250 \
    --network-plugin azure \
    --vnet-subnet-id /subscriptions/<subscriptionId>/resourceGroups/<resourceGroupName>/providers/Microsoft.Network/virtualNetworks/<vnetName>/subnets/nodesubnet \
    --pod-subnet-id /subscriptions/<subscriptionId>/resourceGroups/<resourceGroupName>/providers/Microsoft.Network/virtualNetworks/<vnetName>/subnets/podsubnet \
    --network-dataplane cilium \
    --generate-ssh-keys

Frequently asked questions

• Can I customize Cilium configuration?

  No, AKS manages the Cilium configuration and it can't be modified. We recommend that customers who require more control use AKS BYO CNI and install Cilium manually.

• Can I use CiliumNetworkPolicy custom resources instead of Kubernetes NetworkPolicy resources?

  CiliumNetworkPolicy custom resources are partially supported. Customers may use FQDN filtering as part of the Advanced Container Networking Services feature bundle.

  This CiliumNetworkPolicy example demonstrates a sample matching pattern for services that match the specified label.

  apiVersion: "cilium.io/v2"
  kind: CiliumNetworkPolicy
  metadata:
    name: "example-fqdn"
  spec:
    endpointSelector:
      matchLabels:
        foo: bar
    egress:
    - toFQDNs:
      - matchPattern: "*.example.com"
  

• Why is traffic being blocked when the NetworkPolicy has an ipBlock that allows the IP address?

  A limitation of Azure CNI Powered by Cilium is that a NetworkPolicy's ipBlock cannot select pod or node IPs.

  For example, this NetworkPolicy has an ipBlock that allows all egress to 0.0.0.0/0:

  apiVersion: networking.k8s.io/v1
  kind: NetworkPolicy
  metadata:
    name: example-ipblock
  spec:
    podSelector: {}
    policyTypes:
      - Egress
    egress:
      - to:
        - ipBlock:
            cidr: 0.0.0.0/0 # This will still block pod and node IPs.
  

  However, when this NetworkPolicy is applied, Cilium will block egress to pod and node IPs even though the IPs are within the ipBlock CIDR.

  As a workaround, you can add namespaceSelector and podSelector to select pods. The example below selects all pods in all namespaces:

  apiVersion: networking.k8s.io/v1
  kind: NetworkPolicy
  metadata:
    name: example-ipblock
  spec:
    podSelector: {}
    policyTypes:
      - Egress
    egress:
      - to:
        - ipBlock:
            cidr: 0.0.0.0/0
        - namespaceSelector: {}
        - podSelector: {}
  

  Note

  It is not currently possible to specify a NetworkPolicy with an ipBlock to allow traffic to node IPs.

• Does AKS configure CPU or memory limits on the Cilium daemonset?

  No, AKS doesn't configure CPU or memory limits on the Cilium daemonset because Cilium is a critical system component for pod networking and network policy enforcement.

• Does Azure CNI powered by Cilium use Kube-Proxy?

  No, AKS clusters created with network dataplane as Cilium don't use Kube-Proxy. If the AKS clusters are on Azure CNI Overlay or Azure CNI with dynamic IP allocation and are upgraded to AKS clusters running Azure CNI powered by Cilium, new nodes workloads are created without kube-proxy. Older workloads are also migrated to run without kube-proxy as a part of this upgrade process.

Next steps

Learn more about networking in AKS in the following articles:

• Upgrade Azure Kubernetes Service (AKS) IPAM modes and Dataplane Technology.

• Use a static IP address with the Azure Kubernetes Service (AKS) load balancer

• Use an internal load balancer with Azure Container Service (AKS)

• Create a basic ingress controller with external network connectivity