Power BI Report Server vs. ClickHouse

[vonschultz666]

Hi everyone,

I need help regarding some architectural aspects of our Data Warehouse (DWH) as we are planning to build our warehouse from the ground up. Currently, we are undecided between Apache Superset and MS Power BI for the visualization tool. In this context, we have doubts about the consumption layer.

Our architecture will include three zones: staging, core, and data marts. In the case of Power BI, should we opt for direct query for data ingestion (equivalent to Superset's approach) or should we use import? In either case, where should measures be calculated, perhaps not event in the data mart layer, but rather some separate layer?

Any information, further instructions, or literature recommendations would be greatly appreciated.

[vonschultz666]

For example, if we decide to use direct query mode we would calculate our measures on Clickhouse, similar when using SSAS tabular cubes.

[lbendlin]

In either case, where should measures be calculated, perhaps not event in the data mart layer, but rather some separate layer?

Not sure what your definition of measures is but in the Power BI world they are computed at the UX layer - when users interact with reports.