Dear Anonymous,
we do not have any DWH, so our idea is to:
- Collect raw data into a Datamart as ingestion layer without strong ETL processing (only renaming) - Let's call this datamart as "DATAMART 1"
Once we have all raw data into DATAMART 1 we should be able to develop strong ETL processes by using SQL features and engine into another datamart (DATAMART 2)
By foldable transformation I mean that all queries from DATAMART 1 to 2 are totally foldable, so no additional steps outside the query folding process.
Unfortunately right now simples queries as group-by, join etc... works in the visual query designer but once we try to replicate the exact query into a different datamart we face this Timeout error or the process (if working) tooks hours instead of seconds.
The same issues is experienced if, instead of Datamart 2 we try the same process into a Dataset.