Using stage-marc-import.pl to import bibliographic or author records is really slow. I ran some profiling program and it showed the database calls as the bottle neck. Maybe we should think a better way of doing those, if possible.
Did any of the recent work on imports (using background job etc.) cover this by chance?