Datetime slicer on directquery table from Databricks SQL endpoint

[Tickxit]

Hello everyone,

I am trying to visualise a very large table (near-second data for 20k sensors for a few years). I have it connected to databricks through an SQL warehouse endpoint. The table is called telemetry_augmented and uses directquery mode.

The table is partitioned on year, month, day, and filterting like > datetime or < datetime works well.

On top of selecting the sensors with a slicer (no problem), a between slicer for timestamps is needed, e.g. select data between 2022-01-01 and 2022-01-07.

When placing the slicer on the datetime field of that table (called dp_timestamp), the slicer just won't load. In the SQL warehouse, I see that for this slicer, the MIN and MAX dp_timestamp are queried, which takes over 10 minutes.

This is only for loading the slicers default min and max. I would be fine with a hard coded start and enddate to start the slicer, but I can't find out how to do that. 

I have tried linking this to a datetable to set the slicer based on the datetable, but this resulted in weird behavior too, see my other post for that: https://community.fabric.microsoft.com/t5/Desktop/Filtering-on-linked-date-column-only-keeps-timesta... 

We have also tried adding in databricks another column that indicates in which 10 minute interval it is located. In powerbi we have an imported table, with a column that contains all the possible 10 minute values. Then the two tables are linked and the slicer uses the imported table. However, the query that gets fired to the SQL warehouse is unfortunately not filtering between the timestamps. It looks something like this:

Of course, this is also not a query that is performant.

Any other ideas are welcomed. Thank you.