Re: [Resolverless-dns] Paper on Resolver-less DNS

[Vittorio Bertola <[e-mail]>]

> Il 23 agosto 2019 11:37 Erik Sy <[e-mail]> ha scritto:
> 
> In my view, DANE TLSA is an approach to do public key pinning. The web
> already collected a lot of experience with public key pinning as this
> can be also done using other protocols than DNS. In summary, the Chrome
> browser deprecated this mechanism last year [1]. Assuming that DANE TLSA
> would be widely supported on client systems, I think that Chrome would
> not start using this mechanism to do public key pinning.
> 
> 1:
> https://groups.google.com/a/chromium.org/forum/#!msg/blink-dev/he9tr7p3rZ8/eNMwKPmUBAAJ

According to the message you linked to, the main reason for deprecating HPKP was that it was hard for site operators to make sure that the pinned certificate matched the set of CAs that would be available in browsers at any time in the future, and the chain of intermediate certificates that each CA could use. This is not a problem with DANE, since DANE validates the certificate for the current connection and not for the future (so it's not really pinning it, just validating it every time), and does not depend on CAs and on the browser's acceptance of them. Even "hostile pinning" is much harder with DANE, since it would require the attacker to gain control of both the web server and the authoritative name server, rather than just the web server; and it would cease to have effects immediately once found and remediated.

Actually, moving to DANE validation in the browser would have been quite a reasonable consequence of the analysis in the message.

-- 
 
Vittorio Bertola | Head of Policy & Innovation, Open-Xchange
[e-mail] 
Office @ Via Treviso 12, 10144 Torino, Italy