At 05:09 AM 6/22/2005 -0400, John Leslie wrote:
David Woodhouse <dwmw2(_at_)infradead(_dot_)org> wrote:
On Tue, 2005-06-21 at 23:28 +0100, Tony Finch wrote:
I think we are reaching agreement that only one record should be
published for each SMTP client host name
There's certainly agreement that there should be a limit which is made
explicit in the spec.
I'm going to try declaring consensus on this: that there should be
a limit -- possibly one, but no greater than five -- to the number of
CSA SRV records which may be published at the same version level; and
that the spec should be revised to declare such a limit.
So far, the greatest support has been to limit it to one.
However, David Woodhouse should be given an opportunity to show the
potential benefits of increasing that limit to two.
(I'm not sure we're yet agreed on whether there's a way to avoid an
additional round-trip to look up Address records returned if there are
multiple SRV records at the same version level: regardless, I think we
need to concentrate on the _benefits_ of multiple SRVs.)
Here is a quick survey of how some of the larger email domains set up their
MX records. Assuming they would want to set up their *mailout* records in
a similar fashion, this kind of data might provide some guidance. The MX
column is the number of MX records for the domain, and A/MX is the number
of A records for each MX name. I haven't seen any that exceed 8, but this
is not a big survey. If someone wants to continue, you can find the
biggest mail serving domains at senderbase.org.
domain MX A/MX
rr.com 6 6-7
yahoo.com 4 3-4
aol.com 4 4-5
hotmail.com 4 4
comcast.net 2 2
verizon.net 1 1
charter.com 1 8
--
Dave
************************************************************ *
* David MacQuigg, PhD email: david_macquigg at yahoo.com * *
* IC Design Engineer phone: USA 520-721-4583 * * *
* Analog Design Methodologies * * *
* 9320 East Mikelyn Lane * * *
* VRS Consulting, P.C. Tucson, Arizona 85710 *
************************************************************ *