Re: Devilish: Forget about DNS

Yakov Shafranovich <[email protected]> Mon, 09 Feb 2004 15:46:05 -0500
Newsgroups gmane.ietf.asrg.smtpverify
Organization SolidMatrix Technologies, Inc.
Message-ID <[email protected]>
Hadmut Danisch wrote:

> On Mon, Feb 09, 2004 at 03:06:40PM -0500, Yakov Shafranovich wrote:
> 
>>Because identity #1 can be  successfully verified using DNS. Case in 
>>point - lets take a look at the XMPP protocol, section 14.3 
>>(http://www.ietf.org/internet-drafts/draft-ietf-xmpp-core-22.txt):
...
> 
> Sorry Yakov, but you confused a few things here. The SRV record
> here is something like the MX record, because this is Jabber. 
> As far as I can remember, it works like this: 
> 
> When I, [email protected] wish to chat with you,
> [email protected], I do login into my own Jabber server of 
> danisch.de and send a call to the server of solidmatrix.com, informing
> it that [email protected] wishes to chat with you.
> 
> Then your server is fetching the SRV record for danisch.de in order to 
> find my server, and is giving it a callback.
> 

Sorry I am refering to the wrong section, section 14.4 is the right one:

"Inter-domain connections MUST NOT proceed until the DNS hostnames 
asserted by the servers have been resolved"

This matches each of the servers to the domain its claiming to come 
from, NOT a callback. If you remember, the original proposal by Paul 
Vixie proposed a complementary MX record, and the IETF list for the BOF 
is called MX COMP - complementary.

> 
> This is very different from what we do here:
> 
> - It's a notification-callback scheme. We don't want to have this
>   for e-mail, do we? If we did so, we'd go into a very differnt 
>   transport: When I send you an e-mail, your MTA would have to 
>   fetch the MX records for danisch.de and then contact them to 
>   pick up the message waiting for you. Callback is a nice 
>   verification method, but in the early days of ASRG it was
>   exhaustively discussed that we do not want to have this.
> 

Its not callback, see above.

> - It's based on the assumption that there are no more than 1-2
>   servers with static IP addresses for a domain involved, and 
>   that they all listen for the protocol.  This precondition does not
>   apply to e-mail. 

Aha, that's the crucial difference. However, it is feasible that 
multiple Jabber servers would exist per given domain, which would make 
them run into the same problem. However, my point is that whoever 
designed XMPP used DNS as the logical place to store the domain/IP 
relationship.

Yakov
-------
Yakov Shafranovich / asrg <at> shaftek.org
SolidMatrix Technologies, Inc. / research <at> solidmatrix.com
"I ate your Web page. / Forgive me. It was juicy / And tart on my 
tongue." (MIT's 404 Message)
-------