Azure Kubernetes集群：缩放集实例跨子网/可用区部署配置问询

Absolutely, you can configure your Azure Kubernetes Service (AKS) cluster to spread virtual machine scale set (VMSS) nodes across different subnets and availability zones—this is actually a common practice to boost cluster resilience. Let’s break down how to set this up:

1. Deploying Nodes Across Availability Zones #

First, make sure your target Azure region supports availability zones (most major regions like East US, West Europe, and Southeast Asia do). Here’s how to distribute nodes across zones:

• System Node Pool: When creating your AKS cluster, use the --zones parameter to specify multiple zones. For example:

  az aks create --resource-group myResourceGroup --name myAKSCluster --zones 1 2 3 --node-count 3
  

  This will place one node in each zone (assuming you set --node-count to match the number of zones). AKS automatically balances node distribution across the specified zones as you scale the pool.
• User Node Pools: When adding additional node pools, you can also specify zones with the same --zones flag. This lets you tailor different pools to different zones based on workload needs.
• Key Note: Double-check that your chosen VM SKU is available in all the zones you want to use—some SKUs have limited zone availability.

2. Spreading Nodes Across Different Subnets #

AKS doesn’t support a single VMSS (node pool) spanning multiple subnets, but you can achieve cross-subnet distribution by using multiple node pools, each tied to a separate subnet. Here’s the step-by-step:

1. Create a VNet with Multiple Subnets: First, set up your virtual network and add the subnets you need. For example:

   # Create the main VNet
   az network vnet create --resource-group myResourceGroup --name myVNet --address-prefixes 10.0.0.0/8 --subnet-name Subnet-A --subnet-prefix 10.240.0.0/16
   # Add a second subnet
   az network vnet subnet create --resource-group myResourceGroup --vnet-name myVNet --name Subnet-B --address-prefix 10.241.0.0/16
   

2. Deploy AKS with the First Subnet: When creating your cluster, specify the first subnet for the system node pool:

   az aks create --resource-group myResourceGroup --name myAKSCluster --node-count 2 --vnet-subnet-id /subscriptions/YOUR_SUBSCRIPTION_ID/resourceGroups/myResourceGroup/providers/Microsoft.Network/virtualNetworks/myVNet/subnets/Subnet-A
   

3. Add a Node Pool Tied to the Second Subnet: Now add a user node pool that uses the second subnet:

   az aks nodepool add --resource-group myResourceGroup --cluster-name myAKSCluster --name user-pool-b --node-count 2 --vnet-subnet-id /subscriptions/YOUR_SUBSCRIPTION_ID/resourceGroups/myResourceGroup/providers/Microsoft.Network/virtualNetworks/myVNet/subnets/Subnet-B
   

Bonus: Combine Both for Maximum Resilience #

You can mix these two configurations—for example, deploy one node pool in Zone 1 + Subnet A, and another in Zone 2 + Subnet B. This gives you both zone-level fault tolerance and subnet-level network isolation for different workloads.

Important Reminders #

• Ensure subnets within your VNet have proper network connectivity (default VNet settings allow this, but if you’ve modified NSGs, verify that node-to-node traffic is allowed).
• Each node pool’s VMSS is tied to one subnet, so you can’t split a single pool across multiple subnets—multi-pool is the way to go here.

内容的提问来源于stack exchange，提问作者pawel.kalisz