Which AKS network model minimizes VNet IPs and supports Windows?

AZ-305 By islever Editorial Last updated

Short answer

B Recommend Azure CNI Overlay, which assigns pods from a separate pod CIDR so only nodes consume the 172.16.0.0/16 VNet space while supporting Windows node pools and outbound ExpressRoute connectivity to on-premises.

Question

You have multiple on-premises networks.
You have multiple Azure subscriptions. Each subscription contains a virtual network that is assigned an IP address space of 172.16.0.0/16. Each virtual network is connected to the on-premises networks by using ExpressRoute.
You plan to deploy a container orchestration solution that will use multiple Azure Kubernetes Service (AKS) clusters. The clusters will be deployed to the existing virtual networks.
You need to recommend a network configuration for the AKS clusters. The solution must meet the following requirements:
• Minimize the number of IP addresses required on each virtual network.
• Support outbound connectivity to on-premises datacenters.
• Support Windows node pools.
Which AKS network model should you recommend?
  1. Azure CNI Powered by Cilium
  2. Azure CNI Overlay Correct Answer
  3. kubenet
  4. Azure CNI

Key Takeaways

The question tests the boundary between Azure CNI pod IPs taken from the VNet subnet and Azure CNI Overlay pod IPs taken from a separate pod CIDR; the trap is choosing an IP-saving option such as kubenet without checking Windows node pool support.

Azure CNI Overlay is the AKS network model that minimizes virtual network IP consumption by assigning pods addresses from a separate pod CIDR. This page confirms option B and explains why kubenet and traditional Azure CNI fail the stated requirements.

Common mistake: The most plausible wrong answer is C, kubenet, because it also conserves VNet IP addresses; however, kubenet is the legacy AKS networking option and is not supported for Windows node pools, so it fails a hard requirement.

Expert Analysis

Boundary to decide first: VNet-routable pod IPs vs a separate pod CIDR

Traditional Azure CNI (D) allocates every pod an IP address from the node subnet inside the 172.16.0.0/16 VNet. That makes pods directly addressable, but with multiple AKS clusters in existing VNets it consumes VNet addresses rapidly and strains the shared address plan. Azure CNI Overlay (B) is on the other side of that boundary: nodes still take VNet IPs, while pods are assigned from a separate pod CIDR that is not routable on the VNet. The supplied comment makes the right distinction—overlay avoids giving each pod a VNet IP.

Why outbound ExpressRoute still works

The requirement is outbound connectivity to on-premises datacenters, not direct on-premises-to-pod initiation. With Azure CNI Overlay, pod egress to on-premises is SNATed through the node's VNet IP, so ExpressRoute can carry the traffic. The separate pod CIDR therefore does not disqualify option B; if direct inbound pod reachability were required, traditional Azure CNI or an ingress/load-balancer path would be needed.

Windows and the remaining options

Windows node pools rule out kubenet (C), even though kubenet also conserves VNet IPs. Azure CNI Powered by Cilium (A) is a dataplane and network-policy choice rather than the model that minimizes VNet IP consumption here, and it is not the intended fit for Windows pools in this scenario. D fails the minimize-IP requirement. Therefore B is the answer.

Official Reference

Exam Strategy

For AKS networking questions, first separate direct pod routability from outbound-only connectivity. If the stem stresses VNet IP conservation and only outbound access to on-premises, eliminate traditional Azure CNI immediately. Then use a hard platform constraint—here Windows node pools—to eliminate IP-saving options that do not support it, leaving Azure CNI Overlay. Reserve Azure CNI when pods must be directly reachable from the VNet or on-premises without SNAT or an ingress tier.

Frequently Asked Questions

Does Azure CNI Overlay support direct inbound connections from on-premises to pod IPs?

No. Overlay pod IPs come from a separate, non-routable pod CIDR; inbound traffic normally arrives through a LoadBalancer or ingress, while outbound traffic to on-premises is SNATed through the node IP.

Why is kubenet not the answer if it also reduces VNet IP usage?

Kubenet conserves VNet IPs, but AKS does not support it for Windows node pools, and it is the legacy option. The scenario makes Windows node pools a hard requirement, so kubenet fails.

Does Azure CNI Powered by Cilium solve the VNet IP consumption requirement?

Not by itself. It changes the dataplane and policy enforcement; it does not move pod IP allocation to a separate pod CIDR, so traditional Azure CNI-style VNet IP consumption remains the issue.

More AZ-305 FAQ →

Community Votes

B
100%

100% of anonymous learners picked answer B. Votes are pick records left by other test-takers — they are not the verified answer.

Discussion 1 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 1

  1. gpais 👍 1 Answered B
    Azure CNI Overlay. Unlike traditional Azure CNI—which allocates a dedicated IP address directly from your VNet subnet to every single Pod—Azure CNI Overlay hosts Pods on a completely separate private internal address space (pod-cidr). This saves thousands of VNet IP addresses since only the cluster nodes consume actual IPs from your VNet subnet.

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

Practice All AZ-305 Questions

Access 95 questions with complete answers and detailed explanations.

View Full AZ-305 Practice Test →

← Back to AZ-305 Study Guide