Re: [CUWiN] accept FreeNetworks peering agreement?
Stephane Alnet <[email protected]>
| Newsgroups | gmane.network.wireless.cu-wireless |
|---|---|
| Message-ID | <[email protected]> |
Hi Dave, > I think C-U Wireless should get on board FreeNetworks by accepting > their Peering Agreement. Anybody disagree? I must be missing the whole point of FreeNetworks. :( It seems to me that the different networks are not directly interconnected and that they all use the Internet to interconnect. Doing IPv4 tunneling in an NxN fashion is known to not scale properly (difficult to manage, unstable). Secondly, the mention of BGP seems superfluous. BGP can be used to announce peering policies, but this is obviously not the purpose here since the requirements are for "free transit" (which would need to be defined in BGP terms, BTW). BGP can also be used to announce routes in a limited fashion, but if people are going to have to configure point-to-point tunnels and register their RFC1918 address blocks centrally anyway, this is of limited interest. Overall, the fact that FreeNetworks doesn't provide a BGP server to peer with (which could justify the existence of the whole thing, the same way FreeWorldDialup provides a SIP proxy server to register with) seems to indicate that the designers of the project are quite "unclear on the concept" of BGP peering. As a sidenote, having to "register" RFC1918 blocks in a /16 fashion won't get a lot of people onboard so limits over time the scalability & interest of the idea. [Asking people to publicly register private space doesn't give the project a lot of points on my chart either.] More importantly, beyond the technical details, I'm missing how "end-networks" are going to be able to provide any kind of "transit" anyhow. This looks to me more like a "ah-ah, that BGP thing looks really cool"-moment turned into a couple websites, but with little thinking behind it about what interconnecting disparate networks really means and how to build it properly. S. PS: If the folks at FreeNetworks had mentioned things like MPLS I may have been impressed, but not this time. _______________________________________________ CU-Wireless mailing list [email protected] http://lists.cu.groogroo.com/cgi-bin/listinfo/cu-wireless Project Page: http://cuwireless.ucimc.org