On Tue, Jan 8, 2013 at 9:43 AM, Anton Tveretin <[email protected]> wrote:
> Hello All, > As for me, truly peer-to-peer must be based on IP multicast or something > like. So there will not be any single point of failure, no one could stop > the network. > Existing solutions, based on bootstrap nodes, are really not P2P: think of > the bootstrap node as of 'server'. I've seen severeal networks shut down. > Any more ideas? > Regards, Anton Tveretin. > The problem is that carriers, lacking a business model that makes sense for them, have turned off (or crippled) IP multicast on many networks, so the typical "multicast domain" is no larger than the typical "broadcast domain". So, things like mDNS work well, for very small definitions of "work". It's certainly possible to replace the RELOAD bootstrap node with an mDNS or similar discovery process, and I'm expecting to do this in sort of as "phase 2" for RELOAD. One can find alternative ways to do configuration and used self-signed certs, thereby eliminating most of the need for centralized startup. The problem is the chicken-and-egg -- until there enough RELOAD nodes out there that there might be one to discover, we can't get the protocol off the ground without some fixed resources. But I believe many of us have in mind eventually using RELOAD for the sorts of self-organizing (or self-repairing) networks needed by emergency responders, military organizations, packs of wandering motorcycle gangs, and so on.
_______________________________________________ P2PSIP mailing list [email protected] https://www.ietf.org/mailman/listinfo/p2psip
