Can I create an Azure VM using an M-series size with a Windows image?

Question

I’m running a machine learning model on a dataset with over 200 million records, using Python and VS Code on a Windows-based environment.

To handle the memory and storage demands of this workload, I’ve requested and been approved for M-series VM quotas from Microsoft, which provide the required high-memory and storage capacity.

I now want to create a VM using an M-series size (e.g., Standard_M64ms), but I’m unsure which Windows image is compatible or recommended for this VM family.

Could you please confirm which Windows Server images (e.g., 2019, 2022) are officially supported or commonly used with M-series VMs?

Are there any specific configurations or considerations I should keep in mind when deploying a Windows image on M-series?

I’m using the Azure Portal for provisioning. Any guidance or documentation would be greatly appreciated.

Answer 1

Hi Sankeerth,

Thank you for posting your question on the Q&A portal!

Yes, you can create an Azure VM using an M-series size (like Standard_M64ms) with a Windows Server image (2019 or 2022). M-series VMs are designed for high-memory workloads, including machine learning, and are fully compatible with Windows Server.

Recommended Windows Images:

• Windows Server 2019 Datacenter

Windows Server 2022 Datacenter

Both are officially supported and commonly used with M-series VMs.

Key Considerations:

• Ensure your region supports M-series VMs (check Azure VM availability by region).
• For optimal performance, enable Ultra Disk or Premium SSD storage (see Azure Disk documentation).
• If using GPU acceleration, verify NVIDIA drivers are installed (if applicable).
• Azure M-series documentation
• Windows VM deployment best practices

Let me know if you need further assistance!

Best regards,
Alex
P.S. If my answer help to you, please Accept my answer
PPS That is my Answer and not a Comment

Answer 2

Hi Sankeerth Maddula,
Based on your query,
If a VM has more than 64 vCPUs, it must use Windows Server 2016 or a newer version. While the documentation specifically mentions Windows Server 2016, newer versions like Windows Server 2019 and 2022 are also supported. This requirement ensures that large-scale workloads run efficiently, as older Windows versions may struggle with high vCPU counts.
Please refer the below SS & documentation for more information
M-series
SS

(ex: - Standard_M64ms) VM, with exactly 64 vCPUs, does not exceed this threshold, meaning it can run any of these Windows Server versions. However, to upgrades to a larger VM size, like Standard_M128ms with 128 vCPUs, must ensure their operating system is Windows Server 2016 or later.

Additionally, these VM sizes support both Generation 1 and Generation 2 images, with Generation 2 recommended for better performance. Other features include support for premium storage, accelerated networking, and ephemeral OS disks, all contributing to improved speed and efficiency.
Please refer the below SS & document for details information
Generation 2 VM sizes

SS

Key Considerations for Windows Server Deployment on M-Series

VM Generation: Supports Generation 1 & Generation 2 (Gen 2 recommended for better boot performance & security).

Performance & Storage: Supports Premium Storage, Accelerated Networking, Ephemeral OS Disks, & Write Accelerator.

VM Stability & Maintenance: Live Migration restricted; Memory Preserving Updates not supported.

Nested Virtualization & Advanced Features: Supports Nested Virtualization; ideal for large-scale machine learning & database workloads. Let me know if you'd like more details on any specific point!
__

SS

Please let me know if you have any further query ping in comment

If that information is helpful, please click an accept an answer & upvote👍 for other community members