> Discussing <draft-jiang-p2psip-sep-01> is a welcome opportunity to > freshen up on DHT and application layer (SIP) techniques such as: >
Some feautres from SEP are independent from application layer and could work with other proposals than SEP, I think. > 1. Routing around congested or invisible nodes. This happens in > the DHT layer and is amply documented. Please remember we want to > keep the SIP and DHT layers neatly separated. SEP tries to use keepalive message to communicate status of downstream peers and do some more intelligent routing decisions. It may not be supported by all DHT algorithms. But I think it is a interesting issue which will influence the transactions' success rate. > 2. Advertising and retrieval of classes of data such as > resources/services: SIP proxies, NAT traversal helpers that can do > STUN and maybe some flavors of ICE, NICE, D-ICE :-), plus some > relaying or (perish the thought) tunneling. There may be other > such as buddy lists and voice mail. > We would do well in the WG to define data types and classes to > facilitate faster retrieval (how?; range searches for coded data > types?). > These topics have also been treated fairly well in: <draft-baset- > p2psip-p2pp-01> and earlier in <draft-singh-p2p-sip-01>. > Data model is a core structure in P2PSIP system. It is very important. Earlier RELOAD and P2PP also give descriptions on this topic. The service discovery method in SEP could be used to find a general service and could work with the core data model. > Questions: > > - Is storage and retrieval of data classes relevant to SIP > _generic and independent of the various P2PSIP proposals_? I think, we need a independent service discovery method. You know, the essentienc of the P2P system is to make peers serve each other. So the important task for a peer in the overlay is to find peers where he could get help. So I think it is a important topic which should be considered in a base protocol. > > - What are the common concepts in the three I-Ds that can be > adopted by the WG, irrespective of the P2PSIP protocol choice? > _______________________________________________ P2PSIP mailing list [EMAIL PROTECTED] https://www.ietf.org/mailman/listinfo/p2psip
