top of page

How to Plan Your PRTG Probe Sizing for Azure Sensor Pack Deployment

Updated: Jul 8, 2023

The Azure Sensor Pack is a powerful software for monitoring Microsoft Azure resources. Thanks to our native Auto-Discovery and Monitoring Automation, it can quickly onboard thousands of Azure resources to the PRTG monitoring system. 


In cases of large Azure subscriptions with hundreds or even thousands resources, it is advisable to plan ahead and setup the hardware of your PRTG probe. Below are some sizing considerations for monitoring 2000 Azure resources. We strongly recommend to have a dedicated PRTG Probe with enough memory and CPU power as listed below:


  • Supported infrastructure: VMware (6.x), AWS, Azure, Physical server,

  • Operating system: Windows 2012R2, 2016 Server or 2019 Server,

  • CPU: 8 cores or higher,

  • Memory: 16GB,

  • Disk: 80GB

  • Networking: Make sure the VM you are using is configured for maximum network throughout.

For Azure VMs: The networking capability / limitation of the Azure VM that runs the discovery and monitoring, has significant impact on the speed of discovery (and monitoring).

We have seen that sometimes responses from Azure API can take 10sec vs 2sec when the VM is not provided with the right settings on the Azure side as detailed in the articles below:



Install the PRTG Probe software compatible with your PRTG Core version


Follow the Azure Sensor Pack deployment guide to deploy the latest version of the Azure Sensor Pack (4.3.8 as of writing this blog).


Edit the Azure sensor pack INI file via a text editor such as Notepad++. The file is located at EXEXML/AutoMonX/Azure/Automonx_AzureSensor.ini

Modify the following parameters:

  • THREAD_NUMBER=8

  • PREDATA_VALID_SEC=700

  • SERVICE_RESTART_MIN=1440

Important: The thread number should be equal to the number of CPUs (e.g., 2 sockets with 2 cores each = 4 threads).


Modify your settings as needed and restart the AutoMonX Azure Sensor service.


Follow up on the memory consumption and the overall CPU usage.


162 views0 comments

コメント


bottom of page