Data connections are very poor (none) at any data modelling and attribute abstraction away from the data source, especially compared to the Information designer.
We build information links from attributes sourced from the database, and a assign a column name. This is reusable across other links. Also, as database catalogs and schemas are changed, we can remap this in the information data designer to update source mappings.
With data connections, there is no data modelling possible. While building a view, we can rename a column, but that only applies to that view.
If the schema data source changes, there is no remapping possible and every view needs to be manually updated.
Seem no work is being done in either area of information designer to improve the capability, or the data connection that is even more rudimentary.
Thanks for the valuable feedback Dave. We may not be able to combine column elements from different tables with push down queries but we should consider adding a (library) level above data sources and data connections and make them manageable via UIs and APIs. If would be great to be able to push schema updates to a selection of data sources for example.