I don't know if anything is wrong, but many scripts are slow. Obviously we can't tell you why your script is slow without seeing it and even knowing your data, but here are a few general tips:
Create separate QVWs to actually do the loading from the database and store them in QVDs if more than one application will need to access the data.
Strive for optimized QVD loads. To do that, you can only have a single exists() as a qualifier. So to give an example, if you wanted to load all dates for 2009 and 2010, don't just say "where date >= 20090101 and date <= 20101231". Instead, generate all dates in that range first, then do "where exists(date)". If you have multiple qualifiers, it may be worth doing some as an inner join after the optimized load. It is at least worth comparing load times to see.
Only select and load fields you need.
Minimize the number of times you join to or reload tables.
I do not have my cristalbowl polished and thus cannot see your script nor the amount of data you are trying to read. Apart from avoiding synthetic keys and extensive joins it is always wise to monitor the performance of each step - here studying the logfile is of great help and can assist also in evaluating the effectiveness of changes in programming
