Fabric Capacity - How to trace background operation "Background Query"

[AniBanani2020]

Hello,

Question - how can i find where a background operation "Query" is hitting my semantic model? I am seeing significant background cpu usage from pure report consumers as well as powerbi report developers. Is it possible to figure out where this operation is occuring? How can i stop this from happening.

Unfortunately, Microsoft's documentation on operations are fairly limited. I am also unable to see refresh execution details via the api as they are scheduled refreshes.

Background query = "Queries for refreshing tiles and creating report snapshots"

Background
I have an large semantic model used across the enterprise. Analyts/Pbi developers leverage this to make their own thin reports and use as a base for their own semantic model (reference as a DQ). The semantic model uses incremental refresh and contains around 100mm row fact table. The model is also not set up for cache - although models derived from this model via DQ could be however, the population that i am seeing hit my background usage for this "query" operation are only 5-7 people where as the total consumer pool is well into the 100s.

Any advice is welcome. Appreciate the read

[AniBanani2020]

Question still remains

[AniBanani2020]

Update. Found the semantic model that a developer had built on top of this enterprise model. They had brought in more data, used a virtual relationship to join the table and had a report built out using this. Kicker was user had set his model to save cache. When i removed this, i no longer saw the background hit 24 hrs later.