> > No. The terminal (phone) doesn't need to act as a router to allow > > for multiple devices behind it to connect to the cellular > interface. > > You could eventually have multiple serial connections to > the terminal > > each having its own corresponding air interface > connection. So it can > > act as a host (if you're running the IP stack + app on it) or as > > an L2 device (modem) for e.g. laptops behind it. That doesn't mean > > that there's no advantages in making it a router but it's not > > mandatory. Given that implementers need guidelines now, > it's good if we > > only limit the scope to hosts for the moment and go onto > routers once > > the first is complete. > > Karim, > > In the case of the terminal being an L2 device like a modem > and the IP device being a laptop there might be some care. > > For instance, the laptop IPv6 stack might follow the > specifications and > use NUD over that PPP link. > This requires that the 3G devices in the network actually support NUD > over the PPP link.
Agreed and they will do. I wasn't saying that the network won't support NUDs but the host doesn't necessarily have to. This was just to give the reason for the MAY on NA/NSs in a basic cellular host having only a cellular interface. > > While the draft claims it scope is about the cellular hosts > I wouldn't be surprised if some folks read the MAY and conclude > that they don't need to implement it at the first hop IPv6 router. The first hop router (GGSN) specs belong to 3gpp because it is a special router. I was just using the name router for simplification since that is what it looks like from the host's IPv6 stack. The 3gpp specs support NUDs in this first hop router so no worry about that. > > So I wonder if it makes sense to make this more clear in an > IPv6-over-3G > document which would say > hosts MAY implement sending of NUD messages > routers MAY implement sending of NUD messages > routers MUST implement processing NUD messages (unicast > NS messages) > hosts ??? implement processing NUD messages > > Such careful statements would make sense that things would > always interoperate. Not sure if we should get into putting MUST/MAYs on the 3gpp network parts since the GGSN is not simply a cellular router and is specified in 3gpp specs. However I agree that clarification on the behaviour of the upstream "router" from the host's point of view would fit nicely into an IPv6 over 3G spec. /Karim -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------
