Need detailed information on the protocol behind the distributed
FastTrack network used by Grockster and KaZaA.

What happens when clients first try to connect? Do they look for a
pre-determined server or host and, if so, what happens if that server
is down? Do searches query the whole network, or are they broadcast a
certain number of hops (like Gnutella)? And what exactly is the role
of Supernodes? My goal is to figure out how self sufficient the
FastTrack network really is.

Don't want stuff

Hello,

I started by doing a simple search for the Kazaa Fasttrack protocol.
One of the problems though is that the FastTrack protocol is
proprietary so it will be hard to find any specific technical
specifications of the protocol. The protocol proprietary status is
described here:
[ http://homepage.mac.com/hillmanminx/iswipe/status.html ]

“These networks uses the FastTrack protocol, which is very powerful
and scalable. Unfortunately, the protocol is closed and highly
encrypted.”

However, undeterred I hope to be able to answer your questions about
various aspects of the operations of the protocol from a more
generalised aspect.

The following site I found described various elements of the FastTrack
protocol:
[ http://matrix.netsoc.tcd.ie/~neo/4ba2/p2p/#fasttrack ]

First of, a Supernode is a client which is ranked as a first tier
node. These supernodes are essentially clients that have a broadband
connection, DSL or better. Second tier normal nodes are those with
slower connections such as a modem. The client software works out what
type of node you are based on your connection.

Furthermore according to Kazaa themselves a supernodes act like a
local cache of search information.
[ http://www.kazaa.com/en/help/faq/supernodes.htm ]
Nodes in their vicinity upload a file which contains a list of their
respective downloads. This local cache of search information can then
be used by other nodes to search for downloads that they want.

This helps to answer your other questions. When you connect to the
network the client works out if you are a Supernode or not. If you
are, you connect to other supernodes. If you are a normal node you try
and find a supernode to connect to. Using this method of connection
the FastTrack network would consist of a "backbone" of connected
supernodes which should be fast and reliable.

Searching on the network is slightly different to that of Gnutella.
When a person makes a search request this is transmitted to the
Supernode they are connected to. This supernode then broadcast this
search request to all the supernodes it connects to. This continues
until the TTL - Time To Live - number reaches zero.

"This means, that with a TTL of 7 and with an average amount of Nodes
per SuperNode of 10 a search request will search 11 times more nodes
on a FastTrack network than on GNUtella."

When a client first tries and connects to the network if first has to
register with a central server. This is problematic as if the server
is removed from the network then it will stop working. This is what I
am sure the RIAA is trying to do by attempting to shut down KaZaa and
FastTrack. It also allows for complete control of the network should
they wish to introduce subscription fees for example. In part this
also a possible reason why the Morpheus download client was easily
sabotage by KaZaa and force Morpheus to go over to the Gnutella
protocol.
[ http://www.japaninc.net/newsletters/?list=mmw&issue=9 ]

I hope I have been able to answer all of your questions.

Best regards
aziphirael

Search Terms used:

FastTrack Protocol
://www.google.com/search?q=FastTrack+Protocol&hl=en&lr=&ie=UTF-8&oe=UTF8&start=10&sa=N

KaZaA FastTrack Protocols
://www.google.com/search?hl=en&lr=&ie=UTF-8&oe=UTF8&q=KaZaA+FastTrack+Protocols

Wonderful! Thanks.