Many ServiceGraphs use IntegrationHub ETL for mapping data to CI's. Relationships are added and updated reliably. Example is the Meraki ServiceGraph which uses a single IH-ETL flow to process Organization, Network and Device information (all parent -> child).
When creating a IH-ETL flow in a test environment, relationships changes in data (e.g. VM moving to another ESX host) seem to be handled by inserting a new relationship while not changing or removing the previous one. This results in multiple active relationships where there should be just one. This behavior differs from the observed ServiceGraph.
My setup, a CSV data source and a simple IH-ETL flow that loads VMware VM and ESX host and loads the relation if available.
Dataset one imports successfully and creates the expected relation between ESX1 and VM01:
Dataset two however does add the second relationship but not remove the first one:
Why does this add relationship but not update the existing one (or remove it)?
My config:
