[ietf-dkim] 1193 considered harmful

[Barry Leiba]

>>>> Third, as was pointed out, a sender could hash a large body once and
>>>> send it multiple times, possibly saving a lot of time/effort.
> 
> I'm sort of missing why this is an interesting feature. Reusing the
> hash of the body would only help if you were generating multiple 
> signatures.

Or using the same body in multiple messages.  Suppose "Company I", say, 
is sending a (legitimate, opted-into) mass-mailing of a 70 MB video file 
to, say, 200,000 opted-in recipients.  Suppose also that for some reason 
it has to batch these with different headers, so it can't just sign the 
whole message once.  Saving the work of hashing that 70 MB video 
multiple times would be nice.

I don't consider this a compelling reason (because I think most of these 
cases could -- and would -- just send identical messages, and could just 
hash once in either case)... it's just another reason beyond the others, 
which I do find compelling.

 > I suspect that the RSA signing operation overwhelms the
> SHAx cost by a very good bit on your average size of body.

But that doesn't matter, because we're not RSA-signing the body, only 
the hash.  So it's only the overhead of the hashing that matters.

Barry

--
Barry Leiba, Pervasive Computing Technology  (leiba at watson.ibm.com)
http://www.research.ibm.com/people/l/leiba
http://www.research.ibm.com/spam