Re: [TLS] Working Group Last Call for draft-ietf-tls-sslv3-diediedie-00

[[e-mail] (Martin Rex)]

Erik Nygren wrote:
> Is it worth adding in a note to section 4 (Limited Capabilities) of:
> 
> * Server Name Indication to enable virtual hosting of multiple hosts with
> independent certificates on a single IP address
> 
> ?
> 
> I'm not sure if it's critical enough to hold up last-call?  It is another
> major down-side of the lack of extension support in SSLv3 from a
> deploy-ability perspective.


Your notion seems to be based on a urban legend.

SSLv3 (the protocol), in the final, after-the-fact publication
of November 18, 1996 (which the IETF failed to publish for 15 years,
but finally did as https://tools.ietf.org/html/rfc6106 ),
has _the_exact_same_ support for extensions as TLSv1.0.

Implementations of SSLv3, however, do not necessarily make use of
that capability.

SNI has only limited value, and provides an additional attack surface, e.g.
  https://bh.ht.vc/vhost_confusion.pdf


Personally, I'm much more concerned about newer software that breaks
horribly when no TLS extension SNI is present in the ClientHello,
such as Microsoft ADFS 2012.


-Martin