FastTrack's technology has been documented pretty well over the years and if I recall correctly, the client itself includes huge list of "supernodes" IP addresses that it first tries to connect to in order to retrieve the list of other supernodes from the supernode it managed to connect to and then gets the connection across the P2P network up and running. Supernodes are simply users who have enough bandwidth and fast enough computer to act simultaneously as a client and a server and those supernodes are being assigned "on the fly" by the software itself unless you check the box "don't act as supernode" off from your Kazaa/Grokster settings.