Updated 'GPU network tutorial' and 'manage GPU via PowerShell'articles with SME suggestions from Chinmay.

Author: LouisBerner

articles/databox-online/azure-stack-edge-gpu-connect-powershell-interface.md

@@ -7,7 +7,7 @@ author: alkohli
 ms.service: databox
 ms.subservice: edge
 ms.topic: how-to
-ms.date: 10/01/2023
+ms.date: 02/15/2024
 ms.author: alkohli
 ---
 # Manage an Azure Stack Edge Pro GPU device via Windows PowerShell
@@ -129,7 +129,7 @@ After virtual switches are created, you can enable the switches for Kubernetes c
 
 ## Change Kubernetes pod and service subnets
 
-By default, Kubernetes on your Azure Stack Edge device uses subnets 172.27.0.0/16 and 172.28.0.0/16 for pod and service respectively. If these subnets are already in use in your network, then you can run the `Set-HcsKubeClusterNetworkInfo` cmdlet to change these subnets.
+If you're running the **other workloads** option in your environment, by default, Kubernetes on your Azure Stack Edge device uses subnets 172.27.0.0/16 and 172.28.0.0/16 for pod and service respectively. If these subnets are already in use in your network, then you can run the `Set-HcsKubeClusterNetworkInfo` cmdlet to change these subnets.
 
 You want to perform this configuration before you configure compute from the Azure portal as the Kubernetes cluster is created in this step.
 

articles/databox-online/azure-stack-edge-gpu-deploy-configure-network-compute-web-proxy.md

@@ -213,10 +213,12 @@ After the virtual switches are created, you can enable the switches for Kubernet
 1. From the dropdown list, select the virtual switch you want to enable for Kubernetes compute traffic.
 1. Assign **Kubernetes node IPs**. These static IP addresses are for the Kubernetes VMs.  
 
-    For an *n*-node device, a contiguous range of a minimum of *n+1* IPv4 addresses (or more) are provided for the compute VM using the start and end IP addresses. For a 1-node device, provide a minimum of 2 free, contiguous IPv4 addresses. For a two-node cluster, provide a minimum of 3 free, contiguous IPv4 addresses.
+    If you select the **Azure Private MEC solution** or **SAP Digital Manufacturing for Edge Computing or another Microsoft partner** workload option for your environment, you must provide a contiguous range of a minimum of 6 IPv4 addresses (or more) for a 1-node configuration, and 7 IPv4 addresses (or more) for a two-node configuration.
+
+    If you select the **other workloads** option for an *n*-node device, a contiguous range of a minimum of *n+1* IPv4 addresses (or more) are provided for the compute VM using the start and end IP addresses. For a 1-node device, provide a minimum of 2 free, contiguous IPv4 addresses. For a two-node cluster, provide a minimum of 3 free, contiguous IPv4 addresses.
 
     > [!IMPORTANT]
-    > - Kubernetes on Azure Stack Edge uses 172.27.0.0/16 subnet for pod and 172.28.0.0/16 subnet for service. Make sure that these are not in use in your network. For more information, see [Change Kubernetes pod and service subnets](azure-stack-edge-gpu-connect-powershell-interface.md#change-kubernetes-pod-and-service-subnets).
+    > - If you're running **other workloads** in your environment, Kubernetes on Azure Stack Edge uses 172.27.0.0/16 subnet for pod and 172.28.0.0/16 subnet for service. Make sure that these are not in use in your network. For more information, see [Change Kubernetes pod and service subnets](azure-stack-edge-gpu-connect-powershell-interface.md#change-kubernetes-pod-and-service-subnets).
     > - DHCP mode is not supported for Kubernetes node IPs.
 
 1. Assign **Kubernetes external service IPs**. These are also the load-balancing IP addresses. These contiguous IP addresses are for services that you want to expose outside of the Kubernetes cluster and you specify the static IP range depending on the number of services exposed. 
@@ -545,12 +547,14 @@ After the virtual switches are created, you can enable the switches for Kubernet
    ![Screenshot of the Workload selection options on the Kubernetes page of the local UI for two node.](./media/azure-stack-edge-gpu-deploy-configure-network-compute-web-proxy/azure-stack-edge-kubernetes-workload-selection.png)
 
 1. From the dropdown list, select the virtual switch you want to enable for Kubernetes compute traffic.
-1. Assign **Kubernetes node IPs**. These static IP addresses are for the Kubernetes VMs.  
+1. Assign **Kubernetes node IPs**. These static IP addresses are for the Kubernetes VMs. 
+
+    If you select the **Azure Private MEC solution** or **SAP Digital Manufacturing for Edge Computing or another Microsoft partner** workload option for your environment, you must provide a contiguous range of a minimum of 6 IPv4 addresses (or more) for a 1-node configuration, and 7 IPv4 addresses (or more) for a two-node configuration.
 
-    For an *n*-node device, a contiguous range of a minimum of *n+1* IPv4 addresses (or more) are provided for the compute VM using the start and end IP addresses. For a 1-node device, provide a minimum of 2 free, contiguous IPv4 addresses. For a two-node cluster, provide a minimum of 3 free, contiguous IPv4 addresses.
+    If you select the **other workloads** option for an *n*-node device, a contiguous range of a minimum of *n+1* IPv4 addresses (or more) are provided for the compute VM using the start and end IP addresses. For a 1-node device, provide a minimum of 2 free, contiguous IPv4 addresses. For a two-node cluster, provide a minimum of 3 free, contiguous IPv4 addresses.
 
     > [!IMPORTANT]
-    > - Kubernetes on Azure Stack Edge uses 172.27.0.0/16 subnet for pod and 172.28.0.0/16 subnet for service. Make sure that these are not in use in your network. For more information, see [Change Kubernetes pod and service subnets](azure-stack-edge-gpu-connect-powershell-interface.md#change-kubernetes-pod-and-service-subnets).
+    > - If you're running **other workloads** in your environment, Kubernetes on Azure Stack Edge uses 172.27.0.0/16 subnet for pod and 172.28.0.0/16 subnet for service. Make sure that these are not in use in your network. For more information, see [Change Kubernetes pod and service subnets](azure-stack-edge-gpu-connect-powershell-interface.md#change-kubernetes-pod-and-service-subnets).
     > - DHCP mode is not supported for Kubernetes node IPs.
 
 1. Assign **Kubernetes external service IPs**. These are also the load-balancing IP addresses. These contiguous IP addresses are for services that you want to expose outside of the Kubernetes cluster and you specify the static IP range depending on the number of services exposed.