Re: [Cfrg] Request For Opinions

Alex Alten  wrote:
>At 01:04 AM 5/10/2003 +0000, David Wagner wrote:
>>Maybe I misunderstood your suggestion, but this strikes me as bad
>>advice.  Encryption without authentication (e.g., a MAC or a signed
>>hash) often leads to security holes.  Using per-packet keys is no
>>substitute for authentication.
>
>Yes, this is unusual and doesn't fit his requirements.  One design I
>worked on was retrofitting an old protocol and I had to give up using a
>hash (not enough room in the legacy header and I couldn't change
>packet sizes or tunnel). In this case I was more concerned with raw
>privacy than anything else.

That still sounds a bad idea.  For anyone else who might be reading
this, let me innoculate you against a popular misconception: contrary to
what intuition might suggest, encryption without authentication doesn't
necessarily provide privacy.  It sounds counter-intuitive, but it's true.
Trust me.  We've been through this with IPSec, with SSH, and with WEP,
and it just keeps become more and more clear.  Even if all you want is
confidentiality, you still want to use *both* encryption and a MAC.
_______________________________________________
Cfrg mailing list
Cfrg@ietf.org
https://www1.ietf.org/mailman/listinfo/cfrg