The IP Over NBMA (ION) Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] [Fwd: Unicast inscalability of NHRP]
Bert; > > Yes, this is clearly the case. But it comes at a cost. If the WAN is > > big enough to have to use NHRP instead of ATMARP, it is also big enough > > to have to use dynamic routing on both the IP layer and ATM layer. > > Agreed. However, dynamic routing "at the ATM layer" is not seen as a > negative by some folk. Of course. And you don't have to be bothered by IP routing at all. > It all depends what you want to trade off. Using CSRs in place of > routers, as you say, only gives you the advantage, compared with normal > routers a la RFC-1577, of small cells. The IP applications cannot make > use of ATM's own features, right? They can't turn the knobs. They can. Just as RSVP is a copy of Q.2931, IntServ specs are copy of ATM service classes. :-) > The other side of this would be something more like AEREQUIPA, a > different sort of tradeoff. Unlike NHRP, it is a self-consistent approach having it's own applicability. I hope you can pursue that side of goal somewhere without being disturbed by an architectural principle of the Internet as described in RFC 1958 that, at the internetwork layer, IETF concentrates on IP. Masataka Ohta
|
|