On 9 Jan 2007, at 16:58, Alexander Pevzner wrote:
Justin Cormack wrote:
I would be interested in trying to take a P2P protocol through the
IETF policy process.
I think it should be something like BT - quite simple in terms of
implementation, but with a good usage history. Using the whole IETF
stack isnt necessary to get approval, eg using http - a case could be
made for not using http but it might lose out. Of course if the ietf
protocol wasnt interoperable with BT (and the divergent BT-like
protocols) it might not be successful. But having any P2P protocol
ietf approved would be a very good step. It wouldnt help anyone in
the short term.
As far as I know, BT is a file distribution protocol. For the
standardization purposes I think it would be better to distinguish
between P2P transport and file distribution on a top of this
transport.
Transport protocol should provide:
- peer discovery
- NAT traversal
- p2p data transfer - similar for what TCP sockets does for the
normal
network traffic
The following transport features are optional (may be implemented as a
higher-level protocols), but it is very considerable to integrate them
with the core transport protocol:
- authentication and encryption mechanisms, suitable for P2P usage.
- presence indication
- session initiation
Please note also, to the best of my knowledge, there is still no
published NAT traversal scheme, comparable on its quality to, say,
Hamachi. Published NAT traversal mechanisms, like ICE, lacks some
important knowledge, critical for the high-probability NAT traversal.
While it would be nice to divide it up like this, there are no widely
used,
documented and agreed on standards for these. You would end up
inventing them
more or less for the standardization process. bittorrent is more or less
documented and has several interoperable open source implementations
(which
is needed). Also I cant see IETF being interested in NAT traversal - I
seem to remember this was tried before. But standardising BT as a file
distribution protocol still makes sense (like NFS is a standard). And it
at least sets up a precedent and a potential working group for doing P2P
through a standards process.
Justin
_______________________________________________
p2p-hackers mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/p2p-hackers
