Fred Baker wrote: > > On Mar 23, 2009, at 4:04 PM, Keith Moore wrote: > >>> IMHO, an even more appropriate solution would be to drop the datagram >>> and reply "Destination Unreachable", to cause the originating host to do >>> a better job of address selection. >> >> See, now you're asking the host or the app to do the network's job. > > Actually, no. I am asking the host to do it's job. > > RFC 3484, whatever its issues may be, makes it the host's job to select > an address pair that makes sense.
STRONGLY disagree. The title says "Default". That implies that it's not the only way to do address selection. And as a practical matter, an algorithm like that in RFC 3484 that is based entirely on prefix matching can never work well for all cases even for client-server apps and it certainly can't work well for apps that do referrals. 3484 would never had gotten consensus had it claimed to be mandatory, as it was clearly not workable as a general solution. > Among other things, it asks the system opening the session to look at > the list of addresses advertised in DNS That assumes that the "system" is getting its addresses from DNS. > I'm simply suggesting that if the device has the option of hairpinning > and the option of not hairpinning, not hairpinning makes the most sense. The most sense for whom? Certainly not for the app, nor for the user of the app, nor for the vendor of the app who has to get his app working in dysfunctional environments. > Therefore it is reasonable for the network to advise the host to try a > different address pair. Only if you don't care about the network actually supporting applications. Keith _______________________________________________ nat66 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nat66
