On Tue, 13 Mar 2001 13:55:46 PST, Roeland Meyer said:
Improper conclusion based on a false premise. SRS has shown that root-root synchronization can happen. Probably easier than TLD-TLD syncs, because the rate of change is much lower. That's a
Umm... actually, we showed *THAT* can happen when we first had more than one server for '.' and had to propogate the root from place to place. Two roots that are synchronized are essentially one root. We already know how to have lots of registries feed into a common view of the root. The problem starts when one root knows about .com, .net, and .foo, and another knows about .com, .net, and .bar, and a third has .com, .net, and a conflicting view of a different .foo. Now, how did SRS deal with the first root having an SOA for blat.foo that pointed at an NS entry at 128.154.0.1, and the third root had an SOA for blat.foo that pointed at a different NS entry run by a different company at 119.5.254.9? Guess what? you get different answers based on which root you ask. And I'd be vastly surprised if SRS could tell us how to get consistent results from the two roots in this case. -- Valdis Kletnieks Operating Systems Analyst Virginia Tech