Ted, On Wed, 2009-09-02 at 08:05 -0400, Ted Lemon wrote: > The frustrating thing about this discussion is that Shane is right. > Personally I think rdns is useful, but there is no market pressure to > do it right, and doing it wrong isn't that useful.
I think the idea of IP address to name mapping is useful. > Fundamentally the problem is that the ISP owns the rdns delegation, > and they have no reason to set up a system that few of their customers > will even be able to use. I'm not confident this problem is fixable, > but if it is, that is where the trouble lies, and that is what > probably has to be fixed. When I talk to network engineers they love the idea that the reverse DNS is not in the hands of the end user. In fact, they consider it an affront when I propose moving the address-to-name mapping out of the DNS (which they control) into something closer to the user (which they do not). I vaguely recall someone proposing a simple protocol where you ask a host for it's name. I am very fond of this idea. In IPv6, you could even secure it if CGA were used. It is another way to do the same thing - and DNS has properties you could not duplicate (no caching, no chain of trust, it depends on host reachability, and so on). From that point of view, it is bad. But it would accomplish the main uses of reverse DNS (traceroute, ping, and "who"), with a very small fraction of the infrastructural cost. Does it make sense to pursue such a protocol? If so, where would this work best be done? -- Shane _______________________________________________ DNSOP mailing list [email protected] https://www.ietf.org/mailman/listinfo/dnsop
