On Tue, Oct 17, 2006 at 06:58:42AM +0100, Graham Murray wrote:
"Dick St.Peters" <stpeters(_at_)netheaven(_dot_)com> writes:
Keep in mind that in some cases the reason for the DNS query charge is
that there are per-packet or per-byte charges for traffic to/from some
places, mostly relatively remote places involving expensive
trans-oceanic or satellite links.
So would it not be more sensible for domains in such areas to host
their DNS (and web pages etc) somewhere where there is not such a
charge and therefore they will not be charged for someone
interrogating their DNS records?
Not if this means all data needs to go over an expensive link, which
increases the costs not only for the one publishing the web page but
perhaps, and more importantly, also for the potential customer behind
the same expensive link.
If you are on an island, and your customers are as well, most traffic
will be local and cheap. Move your DNS servers to (for instance) the
US, then both you and your customers pay $$$ to access it. Same for
the web server etc.
Best would be if that island had a server that locals access, and
the rest of the world would access the other server(s) on the other
end of the expensive link. But then you'd still have synchronization
costs.
Indeed, in such a case SPF will cost money, as will other protocols
that poll data somehow. DNS LOC comes to mind, as does SMTP call back.
And what about all those network probes, scanning for open proxies...
It doesn't mean spf is or is not opt-in. But true, it is yet another
protocol and it does cost yet another dime.
It's probably best for such people to publish "v=spf1 ?all", with a
high TTL setting. Or something like "v=spf1 ip4:192.0.2.1 -all" if
they do want a record. Please see my rant at the bottom.
Not publishing SPF would mean relying on negative caching. This will
not result in good caching.
Not answering a query probably means more questions. But if one only
pays for answers, it may actually be the best work around for them.
Now here's the promised rant:
For some reason, people like publishing "v=spf1 a mx ptr ?include:this
?include:that ?include:some_isp_they_have_heard_of_but_never_use
ip4:192.0.2.1 ?all"
They combine "advice" from several places, such as the "wizard", stuff
found on this list and/or other people's records. Such a record is a
real pest for people paying per DNS query.
If you do understand what I mean, you are probably one of the people
giving such advice. Make sure you also tell people why this would
be a bad record, why you should not combine "?include..." and "?all".
Why ip4 should be up front and why "a mx ptr" is probably even worse
than using "ptr" at all.
If you don't understand what I mean, you are probably one of these
people publishing such a brain dead record. "spf-help" is that way --->
2c
alex
-------
Sender Policy Framework: http://www.openspf.org/
Archives at http://archives.listbox.com/spf-discuss/current/
To unsubscribe, change your address, or temporarily deactivate your
subscription,
please go to
http://v2.listbox.com/member/?listname=spf-discuss(_at_)v2(_dot_)listbox(_dot_)com