Need to stop the Azure databricks from populating into the server class table from SGC - Azure.

[ThenmozhiA]

We have many databricks entries in the server class table whose names contain some random number and letters. As , we know databricks are short lived, we dont need them in the cmdb. But SGC- Azure considers databricks as virtual machines. I analysed the payload all the databricks have publisher as AzureDatabricks. So i tried to execute a business rule on the Service gragh conenctor for azure staging table, to delete if the data contains 'AzureDatabricks'. But it dint work. Any efficient way to stop the databricks getting populating into the cmdb.

[Hanna_G]

Hi @ThenmozhiA, 

ServiceNow IRE rules are your best bet. 
Here is a ServiceNow support article on how they work: [CMDB - IRE] How the CMDB Identification and Reconciliation Engine works when passing a CI (as paylo...
And the ServiceNow documentation: Configuring CMDB Identification and Reconciliation

[ThenmozhiA]

The azure databricks comes as server from the azure connector, how will i create a rule. Only in the payload i can see the databricks info. I think for the connector we can have to create script in ETL to prevent the data being processed if it contains the value databricks.

[Josh1]

We’re currently using the Azure ServiceGraph Connector. To block Azure DataBricks from being processed by the Identification and Reconciliation Engine (IRE), I added logic to the "before script" of the SG-Azure VM Transformer Definition. You can find this by searching for "Robust Import Set Transformer", opening SG-Azure VMs, and then accessing the SG-Azure VMs Transformer Definition.

Our cloud team consistently places Azure DataBricks resources under subscriptions that include the keyword "databricks". Using this pattern, I was able to inspect the object_id field in cmdb_ci_vm_instance records, which contains the subscription name, and discard any payloads that include the keyword. This effectively filtered out related records such as:

• cmdb_ci_server
• cmdb_ci_linux_server
• cmdb_ci_nic
• cmdb_ci_os_template
• cmdb_ci_resource_group

This works because every DataBricks payload appears to include a cmdb_ci_vm_instance record.

If your cloud team does not use the "databricks" keyword in subscription names, you’ll need to identify an alternative method to detect Azure DataBricks resources.

When the script detects the keyword, it sets the payload status to "SKIPPED" and assigns a payload reason like:
"Skipped IRE processing of this payload."