Hey, there is an method for designing databases the best way. Maybe you should take a look at that cause it causes the most efficient way of datahandling. Check this :D Product table: 1 | Cookies | 4$ 2 | Milk | 50ct 3 | Potatoes | 4$ 4 | Cola | 50ct 5 | Water | 50ct 6 | Beer | 50ct As you can see, there is a lot of the same info in it (50ct). So a database like the next one would be better: Product table: 1 | Cookies 2 | Milk 3 | Potatoes 4 | Cola 5 | Water 6 | Beer ProductPrices table: 50ct | 2,4,5,6 4$ | 1,3 I know you do not design a database-structure but maybe it helps a hand anyway. Make sure there is not much 'the same text' in the file. But dont make it unreadable either :D. Maybe trying to style the xml like a ftp-directory is the best way? ======================== be skynet ftp.skynet.be ftp http pub/ftp.opensuse.org/opensuse/distribution/ 10.0 i386 delta-iso inst_source i586 delta-iso inst_source 10.1 i386 delta-iso inst_source i586 delta-iso inst_source packman.mirrors.skynet.be ftp http pub/packman i386 i586 belnet ........... de gwdg ftp etc