Rather than using incremental loads you could write a server side script to pull your data from the DB and save as a CSV file, incremental loads can become slow the bigger the table.
A summary for you
Server side script pulls data from DB and saves as CSV files.
QVW file loops over CSVs and creates a mirror QVD file.
All of you reports can pull in the QVDs - LOAD * FROM myUsersCsv_*.qvd (qvd);
An example of speed savings:
To rebuild a 35M row x 40 Column User db table into QV via incremental loads - total build 3.5 hours
To rebuild same table into CSV 8 mins, then point 2 of summary 10 mins. - total build 18 minutes
