----- Original Message -----
From: <hadmut(_at_)danisch(_dot_)de>
To: "Sauer, Damon" <Damon(_dot_)Sauer(_at_)BELLSOUTH(_dot_)COM>
Cc: "IETF MARID WG" <ietf-mxcomp(_at_)imc(_dot_)org>
Sent: Wednesday, June 23, 2004 10:28 AM
Subject: Re: XML/SPF colaberation
Has anyone ever made a calculation about how much the
average DNS traffic and storage will grow and how many
DNS servers will need a hardware upgrade instead of just
a software upgrade?
Has anyone ever cared about the fact that DNS records are
internally binary packed and that they usually do not
store as text what we see when calling nslookup/dig?
I don't enough about the DNS server side, but the traffic concerns has been
thee numero uno issue I had with this whole LMAP/DNS concept since day one.
I mean, it was pretty obvious once you got into it:
Remember, the MAJORITY of the mail is spoofed, nearly 60-80% by most
industry standards which matches exactly what we see. So for 60-80% of
your transactions, you will be exerting a high level of DNS activity
resulting in NXDOMAIN or No Records.
When we finished our system, it was still not ready for release because the
overhead was still too high. Once we implicated what I deem an Important
concept recommended and writing in RFC 2821, we reduced the overhead by
nearly 35%! With this and additional reduction items that minimized all DNS
lookup requirements, we finally released the product. Said one linux
developer, "he seen other systems using the same anti-spam features, but
your method is more efficient."
So efficiency is very important people. But having spoken to some people
who seem to have some DNS background, they gave me the idea, this LOOKUP
overhead is a non-issue. I kept scratching my head about this because it
can't see how it can not be an issue.
This whole SPF vs. XML discussion is ridiculous.
This is not a technical discussion.
This is a religious matter.
This is SPF marketing.
I Agree. We need to take our time and do this right.