After a ton of testing on our future production box (NEWPROD) using many different configurations, I have found that I get a consistent 20% performance boost by having 4 separate MDF files (all on PRIMARY to spread the data) rather than a single MDF file.
The way that I moved the necessary data and views from a single file to multiple files was to restore a copy of my database on our SANDBOX server, then use SSIS via Business Intelligence Development Studio and use the Transfer SQL Server Objects Task to move over the necessary tables and views for testing. This evenly populated the 4 data files on NEWPROD and worked just fine.
Should this method be sufficient to do this for the entirety of the database? I was planning to use a Transfer SQL Server Objects Task and basically just set EVERYTHING to copy over (see screenshot). I imagine that that should essentially give me the copy of the database that I'm looking for, and evenly distribute everything over the 4 files.
Is this an acceptable way for me to do this, or is there another preferred method to break a single MDF across multiple MDFs that people use? I just don't want to lose any of the data, indexes, views, stored procedures, etc. Users and whatnot I can honestly move myself.
I saw this (http:/