Merge pull request #2316 from MicrosoftDocs/main

Auto Publish – main to live - 2025-09-22 22:00 UTC

Author: learn-build-service-prod[bot]

articles/virtual-machine-scale-sets/standby-pools-overview.md

@@ -84,7 +84,6 @@ Users are charged based on the resources deployed in the standby pool. For examp
 
 When using standby pools with Virtual Machine Scale Sets, there are specific limitations and considerations to keep in mind:
 
-- **Azure Autoscale**: Standby pools can't be created or attached to a Virtual Machine Scale Set with Azure autoscale enabled. Ensure autoscale is disabled before configuring a standby pool.
 - **Fault Domains**: Standby pools aren't supported with Virtual Machine Scale Sets configured with a fault domain count greater than 1.
 - **Region and Subscription**: The standby pool must be in the same region and subscription as the Virtual Machine Scale Set it's attached to. Cross-region or cross-subscription configurations aren't supported.
 - **Single Standby Pool**: A Virtual Machine Scale Set can only have one standby pool attached at a time. Attaching multiple standby pools to the same scale set is not supported.

articles/virtual-machine-scale-sets/virtual-machine-scale-sets-rolling-upgrade-custom-metrics.md

@@ -83,7 +83,7 @@ The [application health extension](virtual-machine-scale-sets-health-extension.m
           "publisher": "Microsoft.ManagedServices",
           "type": "<ApplicationHealthLinux or ApplicationHealthWindows>",
           "autoUpgradeMinorVersion": true,
-          "typeHandlerVersion": "1.0",
+          "typeHandlerVersion": "2.0",
           "settings": {
             "protocol": "<protocol>",
             "port": <port>,

articles/virtual-machines/capacity-reservation-overview.md

@@ -58,56 +58,31 @@ From this example accumulation of Minutes Not Available, here's the calculation
 - Creating capacity reservations is currently limited to certain VM series and sizes. The compute [Resource SKUs list](/rest/api/compute/resource-skus/list) advertises the set of supported VM sizes.
 - The following VM series support the creation of capacity reservations:
 
-    - Av2 
-    - B
-    - Bpsv2
-    - Bsv2 (Intel) and Basv2 (AMD)
-    - D and Ds series, v2 and newer; AMD and Intel
-    - Dadsv5 and Dadsv6
-    - Dav4 series
-    - Dasv4 and newer
-    - Ddv4 and v5 series
-    - Dds series, v4 and newer
-    - Dlsv5 and newer series
-    - Dldsv5 and newer series
-    - DCsv2 series
-    - DCasv5 and DCadsv5 series
-    - DCesv5 and DCedsv5 series
-    - ECasv5 and ECadsv5 series
-    - ECesv5 and ECedsv5 series
-    - Dplsv5 and newer series
-    - Dps and Dpds series, v5 and newer
-    - Dplds series, v5 and newer
-    - Eps and Epds series, v5 and newer
-    - E series, all versions; AMD and Intel
-    - Eav4 and Easv4 series
-    - Easv5 and Eadsv5 series and newer
-    - Ebdsv5 and Ebsv5 series
-    - Ed and Eds series, v4 and newer
-    - F series, all versions
-    - Fx series
-    - Lsv3 (Intel) and Lasv3 (AMD)
-
-    At VM deployment, you can set a fault domain (FD) count of up to three by using Azure Virtual Machine Scale Sets. A deployment with more than three FDs fails to deploy against a capacity reservation.
-- At VM deployment for the following VM series for capacity reservation, you can set an FD count of one by using Virtual Machine Scale Sets. A deployment with more than one FD fails to deploy against a capacity reservation:
-    - NC-series, v3
-    - NCasT4_v3 series
-    - NCADSA10_v4 series
-    - NC_A100_v4 series
-    - NV-series, v3 and newer
-    - NVadsA10_v5 series
-    - NVads V710_v5 series
-    - NGads V620_v1 series
-    - M-series, v2
-    - M-series, v3
-- Support for the following VM series for capacity reservation is in public preview:
-    - Lsv2
+  Reserved capacity for these VM series supports high availability using multi-zone or fault domains. The fault domain count for a Virtual Machine Scale Set is limited to 3, enforced at deployment.  
+
+  | Type | VM Series |
+  | ----- | ----------- |
+  | General Purpose (Burstable) | B (Intel) <br> Bsv2 (Intel), Basv2 (AMD), Bpsv2 (ARM) |
+  | General Purpose (General use A, D) | Av2 (Intel) <br> D and Ds series, v2 and newer (Intel) <br> Dd and Dds, v4 and newer (Intel, local disk) <br> Dls and Dlds, v5 and newer (Intel, low memory) <br> Da and Das, v4 series and newer (AMD) <br> Dads, v5 and newer (AMD, local disk) <br> Dalsv6 and Daldsv6 (AMD) <br> Dps and Dpds series, v5 and newer (ARM) <br> Dpls and Dplds series, v5 and newer (ARM, low memory) |
+  | General Purpose (General use, high memory, E )| E and Es, all versions (Intel) <br> Ed and Eds series, v4 and newer (Intel, local disk) <br> Ebdsv5 and Ebsv5 series (Intel, block storage) <br> Ea and Eas, v4 and newer (AMD) <br> Eads, v5 series (AMD, local disk) <br> Eps and Epds series, v5 and newer (ARM) |
+  | General Purpose (General use, compute optimized, F) | Fsv2 series (Intel) <br> Fasv6 and Falsv6 series (AMD)  <br> Famsv6 (AMD) <br> FX series (Intel) <br> FXmsv2 (Intel) <br> FXmdsv2 (Intel) |
+  | Storage optimized| Lsv2 (Intel)-Preview <br> Lsv3 (Intel) <br> Lasv3 (AMD) |
+  | Confidential compute | DCsv2 series (Intel) <br> DCasv5 and DCadsv5 series (AMD) <br> DCesv5 and DCedsv5 series (Intel) <br> ECasv5 and ECadsv5 series (AMD) <br> ECesv5 and ECedsv5 series (Intel) |
+
+  Reserved capacity for these VM series supports only multi-zone high availability. Virtual Machine Scale Sets with a fault domain availability construct is not supported. 
+
+  | Type | VM Series |
+  | ----- | ----------- |
+  | GPU-accelerated compute | NC-series, v3 (Intel) <br> NCasT4_v3 series (AMD) <br> NCADSA10_v4 series (AMD) <br> NC_A100_v4 series (AMD) <br> NV-series, v3 and newer (Intel) <br> NVadsA10_v5 series (AMD) <br> NVads V710 v5 series (AMD) <br> NGads V620_v1 series (AMD) |
+  | Memory optimized | Msv2 Medium Memory series (Intel) <br> Mdsv2 Medium Memory series (Intel) <br> Mv2 (Intel) <br>Mdsv3 Medium Memory series (Intel) <br> Msv3 Medium Memory series (Intel) <br> Mbsv3 and Mbdsv3 series (Intel) |
+
 - Support for other VM series isn't currently available:
-    - M series, v1
-    - M series, HM and VHM
-    - ND-series 
-    - Hb-series 
-    - Hc-series 
+  
+  | Type | VM Series |
+  | ----- | ----------- |
+  | GPU-accelerated compute | ND-series <br> Hb-series <br> Hc-series |
+  | Memory optimized | M series, v1 <br> Msv3 HM <br> Mdsv3 HM and Mdsv3 VHM |
+  
 - The following deployment types are supported: 
     - Single VM
     - Virtual Machine Scale Sets with Uniform Orchestration

articles/virtual-machines/sizes/gpu-accelerated/nvv3-series-retirement.md

@@ -16,7 +16,7 @@ ms.date:     06/13/2025
 > [!NOTE]
 > 1-year and 3-year RI purchases for the NVv3-series will end September 30, 2025.  
 
-On September 30, 2026, Microsoft Azure will retire the Standard_NV12s_v3, Standard_NV24s_v3, Standard_NV24ms_v3, Standard_NV32ms_v3, and Standard_NV48s_v3 virtual machines (VMs) in the NVv3-series virtual machines (VMs). To avoid any disruptions to your service, we recommend that you change the VM sizing for your workloads from the current NVv3-series VMs to the newer VM series in the same NV product line. 
+On September 30, 2026, Microsoft Azure will retire the Standard_NV12s_v3, Standard_NV12hs_v3, Standard_NV24s_v3, Standard_NV24ms_v3, Standard_NV32ms_v3, and Standard_NV48s_v3 virtual machines (VMs) in the NVv3-series virtual machines (VMs). To avoid any disruptions to your service, we recommend that you change the VM sizing for your workloads from the current NVv3-series VMs to the newer VM series in the same NV product line. 
 
 Microsoft is recommending the Azure [NVadsA10_v5-series](/azure/virtual-machines/sizes/gpu-accelerated/nvadsa10v5-series?tabs=sizebasic) VMs, which offer greater GPU memory bandwidth per GPU. With the NVadsA10_v5-series VMs, Azure introduces VMs with partial NVIDIA GPUs and each VM instance comes with a GRID license. This license gives you the flexibility to use an NV instance as a virtual workstation for a single user or 25 concurrent users can connect to the VM for a virtual application scenario. These VMs are targeted for GPU accelerated graphics applications, virtual desktops, visualizations, or small AI workloads. 
 

articles/virtual-machines/sizes/gpu-accelerated/nvv4-retirement.md

@@ -19,7 +19,7 @@ ms.date:     06/13/2025
 > [!NOTE]
 > 1-year and 3-year RI purchases for the NVv4-series will end September 30, 2025.  
 
-On September 30, 2026, Microsoft Azure will retire the Standard_NV4as_v4, Standard_NV8as_v4, Standard_NV16as_v4, and Standard_NV32as_v4 virtual machines (VMs) in NVv4-series virtual machines (VMs). To avoid any disruption to your service, we recommend that you change the VM sizing for your workloads from the current NVv4-series VMs to the newer VM series in the same NV product line.
+On September 30, 2026, Microsoft Azure will retire the Standard_NV4as_v4, Standard_NV4ahs_v4, Standard_NV8as_v4, Standard_NV8ahs_v4, Standard_NV16as_v4, Standard_NV16ahs_v4, Standard_NV32as_v4, and Standard_NV32ahs_v4 virtual machines (VMs) in NVv4-series virtual machines (VMs). To avoid any disruption to your service, we recommend that you change the VM sizing for your workloads from the current NVv4-series VMs to the newer VM series in the same NV product line.
 
 Microsoft is recommending the Azure [NVads_V710_v5-series](/azure/virtual-machines/sizes/gpu-accelerated/nvadsv710-v5-series?tabs=sizebasic) VMs, which offer greater GPU memory bandwidth per GPU. The NVads_V710_v5-series VMs take advantage of the AMD Simultaneous Multithreading technology to assign dedicated vCPU threads to each VM and support NVMe for ephemeral local storage capability.