> Keith Moore wrote: > > ... > > but let's not try to make our task even more difficult by > > insisting that apps support ambiguous addresses or addresses > > with inherently limited reachability. > > For one ambiguity and reachability are different concepts, and for two > there is no ambiguity required. It may happen, so we need to provide > ways to remove or deal with it when it does.
for LL as currently defined, ambiguity is part of the equation. another kind of address might provide a way to resolve that ambiguity. > > Or they want to say that it's okay for the network to impose > > a version of "reality" that doesn't adequately support apps. > > The network connectivity is reality. If the app doesn't work in that > reality, it simply doesn't work. networks are designed to support apps. it's not as if apps can tolerate arbitrary amounts of delay, loss, jitter, etc. providing sane addressing is just another aspect of good network design. > Expecting the network to be globally > accessable and flat is not reality. the network has never been flat in reality, but part of the purpose of IP has always been to provide the illusion of a flat network. we're also a long way from any time when it could be expected that every host in the network could access any other host in the network, but nobody's asking for that now (and I wish you'd stop claiming that people were). > > Look - the whole idea of IP is that the apps don't concern > > themselves with routing, and the network doesn't concern > > itself with data. trying to take away this separation of > > function doesn't make the network more flexible, it makes it > > less flexible. And it doesn't allow the network to support > > more apps, it reduces the number of apps that the network can > > support. It balkanizes the network. > > Balkanization is derived from both policy and technology and has > happened even with the separation. just because it has happened to some degree is not a justification for making it worse. > Allowing the apps to have a hint > about what is happening down below is not going to make a difference > to that. yes, it can - because it's a slippery slope from saying "here's a hint" to "it's okay to do this". and the hints aren't of much use to the apps unless it's to say "it's okay to ignore these addresses". what you're arguing is that it's not okay for apps to ignore these addresses - and that makes the hints irrelevant. the network has a job to do. there's a separation of function between the network and apps inherent in the end-to-end model. if the network fails to do its job, it's not as if the apps are in a good position to take over. what you end up with is a lot of complexity and guesswork on the part of the apps in order to get less functionality, less reliability, more slowly. Keith -------------------------------------------------------------------- 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] --------------------------------------------------------------------
