An administrator is designing a new Azure Kubernetes Service (AKS) cluster in a spoke virtual network (VNet) with the address space . The cluster must scale up to nodes, with each node hosting up to pods simultaneously. The network security team mandates the use of Azure Network Policies for pod traffic enforcement. You need to select the network configurations that will support this scale and meet the security requirements while preventing IP address exhaustion in the VNet. Which two configurations should you implement? (Select two.)
- Configure the cluster network plugin to use Azure CNI Overlay.Cevap
- Define a pod CIDR block, such as , that does not overlap with the virtual network or any peered networks.Cevap
- CConfigure the cluster network plugin to use Kubenet and assign a custom Route Table to the node subnet.
- DConfigure standard Azure CNI with dynamic IP allocation and delegate a dedicated pod subnet within the range.
- EConfigure standard Azure CNI and set the maximum pods per node parameter () to .
Cevap
Configure the cluster network plugin to use Azure CNI Overlay, and define a pod CIDR block, such as , that does not overlap with the virtual network or any peered networks.
The correct configurations involve using the Azure CNI Overlay network plugin and defining a non-overlapping pod CIDR block. Azure CNI Overlay resolves VNet IP exhaustion by only assigning VNet IP addresses to the nodes, while pods are assigned IPs from a private overlay network. This allows a small virtual network range like to easily support nodes and their pods. Additionally, Azure CNI Overlay natively supports Azure Network Policies, satisfying the security mandate. A distinct, non-overlapping pod CIDR block (like ) must be defined to prevent routing conflicts between the overlay network and the underlying VNet.
Adım Adım Çözüm
Anahtar Kavram
Azure CNI Overlay allows AKS pods to be deployed on a private CIDR range to conserve virtual network IP addresses, while still supporting Azure Network Policies and high-performance routing.