Internic, et al.

Tim Wolfe wrote:

Just a thought. The next time gtld servers start puking and Nic starts
playing "we'll get around to fixing it in the next server update" why don't
all the ISPs start having their users send all of their complaints to Won't help much but it sure would feel good...

It would seem to me, and I would have done it this way from the beginning
since I already do it this way with configuration data transferred between
my own servers, that a good way to ensure some reasonable stability on the
various root/gtld servers is for a complete set of backup data to be stored
making it possible to step back one day until a new set of data can be
pulled down and validated. OTOH, I guess maybe the overall stability of
the roots, and the redundancy of the several of them, has made that not
much of an issue in the past. But this isn't the first time the incoming
data itself is bad, and I would venture to guess it is far from the last.

In my automated configuration transfers, my scripts also double check the
data coming in. If the number of items added or deleted exceeds some
configurable value like 10%, then the update is rejected and I get e-mail
(later I might have it page me). It runs on the old data in the interim
and the 10 update cycles that saw valid differences are archived.

I don't know how that would fit into the root servers or even which server
or servers it should be in.

I'm also curious if the data received is cryptographically signed and