[107520] in Cypherpunks
Re: What good's a faster pk algorithm?
daemon@ATHENA.MIT.EDU (Jim Burnes - Denver)
Fri Jan 15 13:56:53 1999
Date: Fri, 15 Jan 1999 11:36:48 -0700 (MST)
From: Jim Burnes - Denver <jim.burnes@ssds.com>
To: Jim Gillogly <jim@acm.org>
cc: cypherpunks@cyberpass.net
In-Reply-To: <369F7644.4B5376B7@acm.org>
Reply-To: Jim Burnes - Denver <jim.burnes@ssds.com>
On Fri, 15 Jan 1999, Jim Gillogly wrote:
> Date: Fri, 15 Jan 1999 09:09:24 -0800
> From: Jim Gillogly <jim@acm.org>
> To: cypherpunks@cyberpass.net
> Subject: What good's a faster pk algorithm?
>
> Bennett Haselton writes:
> > How much does a faster public-key encryption algorithm gain? I thought the
> > conventional wisdom was that encrypted communications involved generating a
> > one-time session key that was enciphered using public-key encryption, and
> > then the rest of the transmission was encrypted with that session key as
> > the secret key -- so that the speed of the secret-key algorithm was what
> > made most of the difference.
>
> If the public-key algorithm were fast enough, the symmetric algorithm
> wouldn't be needed. This would simplify the total system dramatically,
> which would help with analysis of the system and related protocols.
> To paraphrase Einstein, a cryptosystem should be as simple as possible,
> but no simpler.
>
It would also dramatically reduce the authentication/connection
time (and CPU overhead) for SSL and SSH. I haven't seen the
order of efficiency benchmarkes for pubkey vs symmetric key
systems lateley --- blowfish vs RSA or DH.
I imagine it might make a significant difference in latency
for web server connects using both server and client-side certs.
but i'm sure someone out there is much more familiar with
this problem.
jim
