Re: [TLS] WGLC for "Deprecating TLSv1.0 and TLSv1.1"

[Hubert Kario <[e-mail]>]

I think it's fine as it is. This use case is very specific and the impact for 
it is limited. So yes, I think that the delay between publishing and now will 
change the situation even more out of favour of TLS 1.0.

On Friday, 26 April 2019 03:30:00 CEST Viktor Dukhovni wrote:
> > On Apr 12, 2019, at 7:28 PM, Christopher Wood <[e-mail]> wrote:
> > 
> > This is the working group last call for the "Deprecating TLSv1.0 and 
TLSv1.1” draft available at:
> >    https://datatracker.ietf.org/doc/draft-ietf-tls-oldversions-deprecate/
> > 
> > Please review the document and send your comments to the list by April 26,
> > 2019.
> My concern is whether the time is yet nigh for TLS 1.0 to be disabled
> in opportunistic TLS in SMTP, or whether TLS 1.0 remains sufficiently
> common to cause deprecation to do more harm than good via unnecessary
> downgrades to cleartext.
> 
> I don't have survey numbers for SMTP TLS protocol versions across MTAs
> generally to shed light on this, perhaps someone does.  What I do have
> is numbers for those MTAs (not a representative sample) that have DANE
> TLSA records (so presumably a greater focus on security).
> 
> The observed version frequencies are approximately:
> 
> 	TLS 1.0:  1%
> 	TLS 1.1:  0%
> 	TLS 1.2: 87%
> 	TLS 1.3: 12%
> 
> essentially regardless of whether I deduplicate by name, IP or name and IP.
> The respective sample sizes are 5435, 6938 and 7959.
> 
> So if a DANE-enabled sender were to disable TLS 1.0 today, approximately
> 1% of the destination MX hosts would be broken and need remediation.  These
> handle just of 189 mostly small SOHO domains out of the ~1.1 million total
> DANE SMTP domains, but four handle enough email to show up on the Gmail
> SMTP transparency report:
> 
>   tu-darmstadt.de
>   t-2.net
>   t-2.com
>   t-2.si
> 
> So on the whole, the draft should proceed, but some caution may be
> appropriate outside the browser space, before operators start switching off
> TLS 1.0 support.
> 
> I don't see an operational considerations section.  Nor much discussion of
> "less mainstream" (than Web browser) TLS application protocols.  Would a few
> words of caution be appropriate, or is it expected that by the time the RFC
> starts to change operator behaviour the "market share" of TLS 1.0 will be
> substantially lower than I see today even with SMTP, XMPP, NTTP and the
> like.
> 
> [ I would speculate that TLS 1.0's share is noticeably higher among MTAs
>   generally than among the bleeding-edge MTAs that have published DANE TLSA
>   RRs. ]


-- 
Regards,
Hubert Kario
Senior Quality Engineer, QE BaseOS Security team
Web: www.cz.redhat.com
Red Hat Czech s.r.o., Purkyňova 115, 612 00  Brno, Czech Republic