I think the problem is that many protocols are at the wrong level of
abstraction to mandate use of any security controls.

For example, consider IPSEC which at one time was mandatory to implement in
IPv6 but isn't any more because most protocols use SSL rather than IPSEC in
any case.

Should TLS be mandatory for SMTP? Well probably but what if an
implementation uses IPSEC?

The same problem comes up with SSL and HTTP. We can mandate the use of SSL
but not a mechanism to validate the certs and without that SSL has little
value.

We can mandate the use of DNSSEC but that would be counterproductive at
this point as DNSSEC is still a protocol with real deployment problems and
issues that have to be fixed before it is ready. They might have been
addressed earlier if the people involved were not so confident that
deployment was inevitable, making use mandatory would further discourage
fixing issues such as how clients get access to the DNSSEC signatures so
they can do validation.
_______________________________________________
perpass mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/perpass

Reply via email to