Reading various articles in the literature have given me only a few decent methods of keeping track of nodes before->after load-balancing them on a very large network.
One popular method uses virtual-servers and a DHT (Distributed Hash Table)
The problem with DHTs is they are very difficult to maintain, and can result in single point of failure.
Recently I've been looking at implementing a tree of counting bloom filters for finding nodes once their position has been moved.
Can you recommend an alternate method of keeping track of nodes?
 Frank Dabek, M. Frans Kaashoek, David Karger, Robert Morris, and Ion Stoica. Wide-area cooperative storage with CFS. SIGOPS Oper. Syst. Rev., 35(5):202215, October 2001.