Thanks - that's some great info to chew on! (I'm only seeing 1 link..?)

However, it doesn't quite seem to answer my question on the mechanics of how it maximizes storage on the main data page - i.e., at what point does SQL Server decide to overflow, does it do any column pre-sorting or post-sorting once an overflow column is identified (to maximize columns on the main data page)... and the link doesn't seem to offer a means of capturing the overflow column names beyond potentially parsing the memory dump from dbcc page...