qualify rank() over (partition) - question - response (5) by dnoeth

Did you try the query i posted, this should work as expected.
But performnace on a 5 billion rows table will be probably poor when you actually need all those 100 columns (all data will be put in spool twice). 
It might be better to run the QUALIFY on the neccessary columns and then join back to get the remaining columns.
 
Dieter