Blog Post

Azure High Performance Computing (HPC) Blog
4 MIN READ

Comprehensive Nvidia GPU Monitoring for Azure N-Series VMs Using Telegraf with Azure Monitor

vinilv's avatar
vinilv
Icon for Microsoft rankMicrosoft
Sep 30, 2024
In today’s AI and HPC landscapes, GPU monitoring has become essential due to the complexity and high resource demands of these workloads. Effective monitoring ensures that GPUs are utilized optimally...
Updated Oct 16, 2024
Version 8.0
ai infrastructure
hpc
GypJoe's avatar
GypJoe
Copper Contributor
Apr 23, 2025

Very useful thank you!

How do we collect data overtime to analyze performance trends vs real-time monitoring.  I can only find guidance for Telegraf based on Linus VM but I am testing an Azure VM on Windows Server.

 

  • vinilv's avatar
    vinilv
    Icon for Microsoft rankMicrosoft
    Apr 24, 2025

    I've included a section for Windows in the document as well — please take a look. Telegraf can be used on Windows, and I’ve developed a few PowerShell scripts to help with installing and configuring the Telegraf agent for collecting GPU metrics.

    Please note that these steps are part of an unofficial process and should be tested and used at your own discretion.



     

    • GypJoe's avatar
      GypJoe
      Copper Contributor
      Apr 24, 2025

      Yes, I have it set up in my Azure/Windows environment and I can see real-time statistics.  Thank you for that.  However, what do you recommend to use to capture statistics/trends over a period of time?  Thanks again.