Ok, the KPLC monitoring is exposing me to many many crazy scenarios on data collections and sets which I cannot complain as it's extremely interesting. One question:  suppose that I wanted to create a torrent like data inputs where data can be shared amongst the various peers. How do torrents do it via a different ftp algorithm because they don't use regular ftp. 

Anyone with an in-depth understanding of torrent architecture would be able to help provide some clues. This would be appreciated.

Rgds. :-)