The mobile phones implement the P2PP protocol (http://tools.ietf.org/id/draft-baset-p2psip-p2pp-01.txt) and Kademlia DHT. Depending on a situation a mobile phone may act as a peer or a client. If it becomes a client it connects to some of P2PP peers and uses a subset of the P2PP protocol, called a "client protocol" (mainly PublishObject, RemoveObject, LookupObject), to perform PUT/GET operations. If a mobile phone can act as a peer then it uses the full P2PP protocol and runs Kademlia code (the clients do not run DHT).
In addition the demo includes TURN like relays that publish their contact information in the overlay and help clients and peers to traverse NATs/Firewals. Marcin >-----Original Message----- >From: ext Henry Sinnreich [mailto:[EMAIL PROTECTED] >Sent: 22 January, 2008 17:30 >To: Enrico Marocco; Matuszewski Marcin (Nokia-NRC/Helsinki) >Cc: [email protected]; Henning Schulzrinne >Subject: RE: [P2PSIP] HIP: optional, mandatory? > >Enrico, > >>So, if I understand correctly, Marcin's mobile devices >connect directly >>to P2PP peers running on PlanetLab nodes without effectively >being part >>of PlanetLab, establishing an overlay made of both PL and >non-PL nodes, > >The mobile devices run the _client protocol_ that has been the >topic of many discussions on this list. > >Maybe Marcin can give us the reference to the latest >description of the client protocol for P2PP implemented for >this trial. > >Thanks, Henry > >-----Original Message----- >From: Enrico Marocco [mailto:[EMAIL PROTECTED] >Sent: Tuesday, January 22, 2008 6:08 AM >To: Henning Schulzrinne >Cc: [EMAIL PROTECTED]; Henry Sinnreich; [email protected] >Subject: Re: [P2PSIP] HIP: optional, mandatory? > >Henning Schulzrinne wrote: >> PlanetLab is many things, but stable is not the adjective I would >use... >> These are generally servers run on a volunteer basis by graduate >> students in various university labs; they routinely get >disconnected, >> crash or get overloaded with other experiments. >> >> Relative to consumer nodes, they do tend to have public IP addresses. > >So, if I understand correctly, Marcin's mobile devices connect >directly to P2PP peers running on PlanetLab nodes without >effectively being part of PlanetLab, establishing an overlay >made of both PL and non-PL nodes, right? > >-- >Ciao, >Enrico > _______________________________________________ P2PSIP mailing list [email protected] https://www1.ietf.org/mailman/listinfo/p2psip
