On Wednesday, April 16, 2003, at 04:20 PM, Hadmut Danisch wrote:
btw, the second version of the RMX draft ist available
at
http://www.ietf.org/internet-drafts/draft-danisch-dns-rr-smtp-01.txt
Perhaps I'm missing it, but I still don't see a solution to the problem
of "needing more RMX records than can safely fit inside a UDP response
packet".
Is it your plan that $BIG_SITE which might have hundreds of valid
networks around the world should incur on its NS-set, TCP lookups from
MTAs around the world over and over again?
This, to me, seems a significant barrier to entry. The sites who are
most likely to want to use a DS/RMX solution are the ones who have many
many networks, and it would be committing Name Server Suicide to
subject them to what would in the end amount to an
created-to-specification DDoS, because they have to fail over to TCP
for every MTA request.
You've made the comment to me, privately, that you consider this bad
network design, but the reality is that for most applications it's
called "redundancy" and/or "wide customer base", either of which is a
perfectly valid solution for having large, diverse, network address
ranges like that.
Any solution along these lines, IMHO, *has* to be certain that its
responses are going to stay in within the scope of a standard UDP
response packet, or it's going to be a non-starter when it comes to
adoption by the people by whom its adoption is most desired.
D
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg