Network Management

Performance Degradation When Monitoring Azure Cloud Accounts with Databricks VMs in SolarWinds Platform

This article provides information about a performance degradation issue in the SolarWinds Platform when monitoring cloud accounts that include Azure Databricks virtual machines (VMs). The issue causes database performance problems due to the platform incorrectly handling VMs with frequently changing unique identifiers.

First published date

5/9/2025 5:05 PM

Last published date

10/21/2025 7:51 PM

Overview

In environments where Azure cloud accounts contain Databricks virtual machines, users may experience significant performance issues when Auto Monitoring is enabled. Databricks VMs are often ephemeral, meaning they are frequently created and destroyed throughout the day.

 

When Auto Monitoring is turned on, each newly created Databricks VM is discovered as a new unique instance due to the use of changing UUIDs/GUIDs, even though they represent the same logical VM from the Azure environment.

 

The SolarWinds collector continues to poll these new instances, while polling data for previously seen (but now deleted) VMs remains in the queue. Since the associated VM no longer exists in the database, the collector repeatedly attempts to insert the data, creating a loop that impacts overall polling and database performance.

 

This behavior can be verified by reviewing the Events table, where new cloud instances (with unique IDs) are seen being added and removed multiple times per day. Over time, this can severely impact database responsiveness and collector stability.

Product section

Network Performance Monitor

Cause

The issue is caused by how Azure Databricks provisions ephemeral VMs. These VMs:

  • Are instantiated and destroyed frequently throughout the day.
  • Receive new UUIDs/GUIDs with each creation cycle.
  • Appear as new, unrelated devices to SolarWinds Platform each time they’re created.
  • Leave behind orphaned polling data when removed, which the collector continues trying to insert.

Because the SolarWinds Platform is currently not optimized to recognize and de-duplicate these types of resources, the result is database insert failures and collector queuing loops.

Resolution

A fix for this issue is included in SolarWinds Platform 2025.4. Consider upgrading your environment. If you can't upgrade, use the workaround below.

 

Workaround:

To prevent this issue, do not enable Auto Monitoring for cloud accounts with known Azure Databricks VMs. Instead, manually choose the instances/VMs to monitor using the steps below:

 

  1. Navigate to the Cloud Infrastructure Monitoring Settings page.
  2. Click Manage Cloud Accounts.
  3. Select the checkbox for the cloud account containing Databricks resources.
  4. Click Choose Instances/VMs.
  5. Deselect any Instances/VMs that are identified as ephemeral or related to Azure Databricks.
  6. Click Save to apply the changes.

       

      This prevents these ephemeral VMs from being automatically monitored and stops the collector from queuing unusable data.