Yeah well, it seemed the easiest and quickest way to get a hold of a 400k records and toy around with it without wasting more than 5 minutes. And obviously no sane person would do a group by 50 columns in 4 tables query especially knowing beforehand that the records were unique and there's no point grouping anyway :p

MySQL vs SQLServer. I did test that once; a web application, same tables, same data in tables, same queries, same machine; db engines were SQL Server 2000 and whatever mySQL was declared "stable" that morning, I forgot (was a while back). SQL server won hands down, performance-wise. (edit: could be because the mysql odbc drivers were crap though, I haven't been curious enough to play with MySql in a Linux environment)