> -----Original Message-----
> From: [EMAIL PROTECTED] 
> [mailto:[EMAIL PROTECTED] On Behalf Of Martin Schütte
> Sent: Saturday, May 10, 2008 10:33 AM
> To: [email protected]
> Subject: [Syslog] lower requirements? (Re: 
> I-DAction:draft-ietf-syslog-transport-tls-12.txt)
> 
> >    The transport sender (TLS client) has three different options for
> >    authenticating and authorizing the transport receiver 
> (TLS server).
> 
> I do not know if this has been discussed previously, but what 
> is your opinion on lower requirements in order to get 
> transport-tls supported by embedded devices, i.e. switches 
> and printers?
> 
[Joe] TLS is deployed to mitigate certain threats such as those described in 
the document.  If you fail to mitigate these threats then the value of 
deploying TLS is diminished.  

> Scenario:
> I could imagine a printer (as the client) having a 
> self-signed certificate and no ability to authenticate the 
> server's certificate.
> As long as the server has a copy of the client's certificate 
> and can verify it, a secure transport is possible. As an 
> admin I would rather configure this one-way authentication 
> and get a TLS-enabled device than having to fall back to UDP.
> Should this be an allowed scenario to be covered by tls-transport?
> 
[Joe] The scenario above does not mitigate server masquerade.  I'm not sure 
this would be generally acceptable. However, I realize there are systems that 
work much in this way today.  

> Scenario2:
> Say the same printer with its self-signed cert is 
> configurable with a CA-cert that enables it to authenticate 
> the server (but maybe without checking the certificate's 
> CN/dNSName/IP).
> That would allow a reasonably secure setup. -- Should this be 
> an allowed scenario to be covered by tls-transport?
> In my opinion it should be, thus I would like to keep the 
> requirements on authentication rules as simple as possible.
> 
[Joe] This seems like a workable scenario to me in some environments.   This 
would require the CA to issue certificates only to authorized parties.  Perhaps 
the argument can be made to for the configuration of a root as a MUST and the 
specific types of subject Name checks as RECOMMENDED with the appropriate 
security considerations discussion. 

> -- 
> Martin
> _______________________________________________
> Syslog mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/syslog
> 
_______________________________________________
Syslog mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/syslog

Reply via email to