On Oct 27, 2010, at 1:45 PM, Fred Baker wrote: > > On Oct 27, 2010, at 10:04 AM, Chris Engel wrote: > >> I actually wouldn't have any objection to some mechanism built into NAT that >> allowed for a requesting host/application to be informed of the Public >> Address it would be assigned by the NAT device. > > I don't object to that either. I do question the requirement apart from > Dynamic DNS. Keith's comment earlier today indicated that he didn't like the > mechanisms that exist for doing so, and specifically didn't want to have to > send a datagram to another system.
I'm not sure where the latter bit came from. I don't object to having the apps or hosts send a datagram (or open up a tcp connection) to an address in order to obtain a global transport address associated with a connection. (there are cases where a datagram request is better and cases where it makes more sense to have a tcp or similar connection - but I'm also envisioning a more general mechanism that also deals with additional problems caused by stateful NATs.) What I object to is requiring apps that do referrals to have "prime real estate" (in the form of servers that reside in global address space) that exist to allow those apps to do referrals between address realms. That's essentially what ICE requires, and it's an unreasonable burden to impose on all apps that need to do referrals. Keith _______________________________________________ nat66 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nat66
