Re: Devilish: Forget about DNS
Yakov Shafranovich <[email protected]> Mon, 09 Feb 2004 16:02:09 -0500
| Newsgroups | gmane.ietf.asrg.smtpverify |
|---|---|
| Organization | SolidMatrix Technologies, Inc. |
| Message-ID | <[email protected]> |
Mark Foster wrote: > On Mon, Feb 09, 2004 at 02:20:51PM -0500, Yakov Shafranovich wrote: > >>What bothers me is that there are two separate identities at stake here: >>1. Identity of the incoming MTA. >>2. Sender's identity > > Many of the MTAs in use today support the STARTTLS mechanism. In order > to accomplish this, the MTA must be configured with a X.509 certificate > that is either self-signed, or signed (issued) by a certficate authority. > > Also, some of those MTAs can be configured to use the same (or different) > certificate outgoing (as a client), and to recognize certificate authorities > for trust verification. > > But AFAIK none of the MTAs I have looked at currently support CRL or OSCP. > The question is who issues the certificates. If the certificate is self signed than its useless unless you can somehow tie it in to something (domain, IP, etc.). > In talking about secure Internet email, everyone seems focused on S/MIME certs > for end-users. Why not also advocate the use of certificate verification (PKI) > in message relaying MTA-to-MTA? This just seems like a natural direction for > the email infrastructure. > This is exactly what I am wondering - we need to verify two separate things - the MTA and the sender, and the methods to do so will vary. Yakov P.S. Mark, your list delivery has been disabled, contact me off-list. ------- Yakov Shafranovich / asrg <at> shaftek.org SolidMatrix Technologies, Inc. / research <at> solidmatrix.com "Among all our enemies / The ones to be most feared are often the smallest" (Jean de la Fontaine) -------