Hi Vidya, See comments below,
On Tuesday 08 August 2006 22:24, Narayanan, Vidya wrote: > > > > so, the router will not send unsolicited RADV but will only > > answer to router solicitations is that it? > > There will have to be no multicast RAs. Some networks detect the > mobile at L2 and send a unicast RA when the link local address is > available (via L2 means). In others, yes, there will be no > unsolicited RAs. > > > wouldn't this pose some problems with movement detection for > > instance? > > i mean there will no periodic RADV to detect the we are on a > > different link... but i guess you are assuming some for > > netlmm... but this is off topic for the int ml i guess > > Yes - due to this restriction, all nodes attaching to links within > the NETLMM domain will have to follow NETLMM defined behavior - I > don't see how there can be a mix of nodes on the network, some that > obtain NETLMM services and some that don't. Those are actually the > reasons applicability of this model to all sorts of environments > becomes weird. > > It fits the cellular world, where the attached hosts are all paid > subscribers that may uniformly get NETLMM services. In the > enterprise world, where true shared prefixes are used, for > instance, one may want to provide NETLMM services just to the > enterprise hosts and not for guest devices or such. This model will > not really allow for such mixed deployments. There is no requirement that the MN follow a "NetLMM defined behavior", as you put it. The MN is supposed to follow standard ND and IP layer movement detection (DNA) When the MN moves within a NetLMM domain, the IP layer movement detection (DNA) will always conclude that the node hasn't moved *from IP point of view*. The only thing that needs to be updated is the default router. This can be done quickly if the MN implements the DNA protocol. Now if DNA isn't implemented the default router update would be slower because NUD has to be invoked, but that's not specific to NetLMM. It is the case for every moving node which does not implement IP layer movement detection (DNA) IMHO link-layer considerations (e.g. cellular) are orthogonal to that issue. Best, --julien _______________________________________________ Int-area mailing list [email protected] https://www1.ietf.org/mailman/listinfo/int-area
