Hi!
I have a hybrid model DQ/Import with a large fact table that is in Direct Query for the data source I'm using databricks SQL warehouse.
I've created aggregation tables and configured the aggregations to respond to the most used query patterns and it's working well. But even if a report doesn't need to go to DQ I still see queries in databricks like:
Listing cross references
Listing primary keys
Listing columns 'catalog: ...
That are taking time and slowing the report.
One other major issue is queries that go to DQ that take ~20seg or less to run and return only few rows are causing an increase in the CU% and sometimes cause throttling.
Why are DQ queries consuming so much capacity units and what are those "Listing..." queries? Is there a way to improve this behaviour?
This way it's not possible for the report to be used by multiple users.
Thanks in advance.