Hi Ed,
The one weakness I see in the presentation of CoDoNS is one that is
common amongst academic exercises. While it treats a technical
problem in a formally defined say, it suffers from the "assume
frictional surfaces" syndrome. This disease is not fatal, it is more
like the flu, meaning that the work is worthwhile and there are some
nuggets of real helpful technology, but taken as one package it is no
better than what is out there today - which does not have to assume
frictionless surfaces.
It's certainly true that operationally oriented folks would like to see
issues addressed at a different level than academic folks. And this is
entirely reasonable; I can see how academic papers or quick email notes
would engender that feeling.
Just to put things in perspective:
- CoDoNS showed how to use a P2P system to serve names, and to do so
with high performance. Previously, this was considered
impossible because DNS query distributions follow Power Law
distributions, and such query distributions were considered
to be impossible to handle. There are lots of papers on
heuristics-based approaches to managing a name cache that
have no guarantee of working, and there is a paper from highly
esteemed colleagues at MIT who wrote that serving DNS with a
P2P substrate was impossible (which was the state of the art
at the time). We showed how to do it.
- We rewrote a new DNS implementation based on this new P2P architecture
that is backwards compatible with existing resolvers and
supports the same namespace as the current DNS. Remember how
long the BIND9 effort took, with extensive resources? CoDoNS is
a comparable effort, plus it has a P2P organization, plus it
has a radically new way of managing cache contents and cache
update traffic. By just 2 people total (actually, almost
entirely 1 person, my former graduate student, Rama).
- We serve some names authoritatively through our system to protect
them from DoS, as evidenced by the www.electoral-vote.com
effort, where we provided both DNS and CDN services to a site
under serious threat of DoS attacks.
So, you can say whatever you like about academic research in general,
but when it comes to the CoDoNS, we know about friction-bearing
surfaces.
Best,
Gun.
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf