For my use cases / environment it’s not so much about technical blocking as it is using the best tool for the job. I have (read only) integrations that CHURN through a boatload of data and they can run in milliseconds on SQL server with stored procs, etc., and sure I can rewrite them to bubble everything up through the application layer but now it takes minutes to run and can slow the app server when the whole thing is really just a complex series of SQL queries with temp tables all over the place.
EDA (and I believe Grow as well) is also using DB scripts to run the heavy duty queries on the SQL server, it’s not bubbling everything up through the application. That is how I’d prefer to keep the data intensive stuff.