Date: Mon, 15 Jul 1996 11:08:11 -0700 (PDT) From: "JULIAN Elischer" <julian@ref.tfs.com> To: dennis@etinc.com (Dennis) Cc: archie@whistle.com, hackers@FreeBSD.org Subject: Re: interfaces, routes, etc. Message-ID: <199607151808.LAA24788@ref.tfs.com> In-Reply-To: <199607151659.MAA20029@etinc.com> from "Dennis" at Jul 15, 96 12:59:50 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > A. Cobbs writes.... > > >Dennis writes: > >It's not a point-to-point link, it's a NBMA link -- a type of link > >of which traditional point-to-point is an easily handled degenerate > >case. we will try follow what the NBMA working group suggests. > > Then the POINTOPOINT flag shouldnt be on. NMBA is not PTP. If you want to > model NMBA, fine, but dont degrade to meaning of one flag by adding another. > You eithe model at NMBA or PTP, but not both simulaneously. The community > generally does not model as NMBA because it is difficult to model > effectively with > existing routing protocols, and it has generally been determined that > virtual PTP > modelling works better and is more efficient. [...] > > You're not understanding what it is. What you have is virtual PTP > connections, not PTP with addressing, and not multi-point either. > Without fully implemented discovery capability > (like ARP on ethernet), you cant effectively use the single interface > model without re-writing a lot of stuff. And the bottom line is that > no one is using it, so you likely won't be very compatible if you do. > [...] > > You can define it however you like, but you wont have a working > product that will properly interface with the outside world. > > [...] > you didnt say that in your message. The confusion is that your flags are wrong. > Its not PTP and its not multicast. You'll break lots of stuff if you do this. > > Please dont do it with these flags....... there is one spare flag bit in the interface flag word. it will mean adding code to check PPP OR NBMA because they are similar in so many ways as far as the routing is concerned. (e.g. they only care about the remote address). > > You'll be sorry to find out how many devices implement frame relay differently, how different can they get? If I set an IP address to match a DLCI in the routing table (aka arp table) then how can that not work? If I set a couple of bits there asside to indicate encapsulation type, I can't see how it can NOT work.. > and that you wont be able to talk to many devices. for example? > > Unfortunately, what you should be doing is revamping aliasiing because > aliasing in its present for is a terrible kludge. Now you're trying to > extend the kludge in another inappropriate way. Now there I agree.. The aliasing needs to be looked at.. It was never designed for half of the things it's being used for..
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199607151808.LAA24788>