We have a couple very large datasets that need publishing out of one SQL ArcGIS Server database into a 'Publishing' database (20 million records).
The data needs to be joined to 3 tables for additional attribution, plus a few aliases etc etc. Overall, they are simple 1-M joins based on traditional pk/fk - no spatial work required, except the input geometry from the large point dataset (basically joining to owner names, addresses in one table and a 'hierarchy' table for System Names and a couple other attributes)
Typically, we prefer to 'pre-process' as much as possible to make our nightly ETL processes more simple and perform better (join + export = slooooow)
Any best practice suggestions for creating views (SDE.Default) while keeping geometry?