ietf
[Top] [All Lists]

Re: https at ietf.org

2013-12-11 00:27:12

On Dec 10, 2013, at 6:00 AM, John C Klensin <john-ietf(_at_)jck(_dot_)com> 
wrote:

While the integrity checks of DNSSEC provide some protection
against some types of attacks on the "data quality" part of the
DNS environment, the attacks they protect against are very
difficult.  An attacker with the resources to apply them would
almost certainly find it easier, less resource-expensive, and
harder to detect to attack registry databases (before data are
entered into DNS zones and signed), registrar practices, or
post-validation servers.  Non-technical attacks, such as the
oft-cited hypothetical NSL, are easily applied at those points
as well -- much more easily than tampering with keys or
signatures.

Dear John,

The opacity of CAs, DNS, and BGP place all forms of security at risk!  These 
mechanisms are never stronger than their weakest link.  Without burdensome 
cryptographic checking, no Internet service should be trusted.  DANE in 
conjunction with DNSSEC affords a much needed transparency to expose exchanges 
at risk.

TLS should be considered a two-way certificate exchange resolving domains 
rather than individuals.  For example, a federated service like email lacking 
two-way certificate checks can not be defended from abuse, nor can privacy of 
those using such a service be assured.  Transparency in the security mechanism 
should fully illuminate domains, not individuals.  TLS and StartTLS contain 
elements for two-way certificate exchange and can make a needed transition from 
CAs to DANE while providing a verifiable chain of trust to the controlling 
domain.

Of course, websites have adopted synthetic domains as an alternative for web 
cookies. When naughty synthetic domains are used, no amount of encryption 
protects individuals when metadata remains fully apparent.  Perpass documents 
should have included stronger statements about protecting services as apposed 
to suggesting shortcuts in the guise of affording privacy.  No conversation 
should ever be considered private without first ensuring the controlling domain 
at each end of the exchange.

Regards,
Douglas Otis

<Prev in Thread] Current Thread [Next in Thread>