[2677] in Kerberos
Re: New Govt. Key Exchange Protocol..
daemon@ATHENA.MIT.EDU (Ganesan)
Mon Apr 19 18:30:57 1993
From: bf4grjc@socrates.MIT.EDU (Ganesan)
To: Chuck.McManis@Eng.Sun.COM (Chuck McManis)
Date: Mon, 19 Apr 1993 18:19:03 -0500 (EDT)
Cc: kerberos@Athena.MIT.EDU
In-Reply-To: <9304192127.AA01891@pepper.Eng.Sun.COM> from "Chuck McManis" at Apr 19, 93 02:27:44 pm
Reply-To: bf4grjc@bell-atl.com
>
> >BTW: I assume that this whole thing is basically focussing on
> >telecommunications. It's fairly irrelveant to users at computers,
> >in this day and age of anonymous terminals.
> This is a fairly naive view.
^^^^^^^^^^
Is it? Lets see:
- the algorithm will NOT be available, so there WILL be no software implment-
ation. Unless you are willing to believe that every piece of computing
hardware will have the clipper chip (completely unrealistic in a five
year planning period), this system is basically unusable for computers.
- Since the algorithm is not being published, this does not seem to be the
"son of DES" people have been awaiting, either. A DES replacement simply
has to be able to work in s/w to make a dent on the computinf industry.
- The security of the chip is BASED on the ability to TIE [User, Serial No.,
Secret Key]. How will this system be conveniently used when users are not
TIED to a chip? i.e. the govt. may win the right to tap the outward line
of user X who uses say an IBM 3090. If the 3090 has ONE CLIPPER chip, the
govt., to tap communications from it will ahve to obtain warrants to tap the
lines of all 2000 (say) users.
- When you buy a computer (or an ordinary phone) you do NOT have to 'register'
it. You can sell it without notifying anyone - a scenario in which this
new proposal is practically useless. When you buy a ceullar phone or a
PCS, you MUST register your self. You CANNOT sell it to someone else, without
deregistering and re-registering (basically like selling a car)- the new
system seems to be designed for JUST this environment.
- This 'system' does NOT do authentication, and does NOT do key exchange, and
in that sense is NOT a replacement for a Kerberos or Diffie/hellman or RSA.
For instance: Describe how you would do PEM using this system between two
corporations?
- In the INTERNET any two users can do key-exchange using public-key (even
the new system requires the use of some external key exchange mechanism)
and do DES or FEAL or something. NOTHING stops them. In the PCS world
a PCS is a fairly non-programmable device, and to do your own encryption you
MUST replace h/w, something that is somewhat harder to do (except for
people who REALLY want to: like terrorists).
I agree that EVENTUALLY the govt. will get around to wanting to snoop (and
I agree that there is valid reason) on the INTERNET, but THIS standard seems
clearly targetted at the PCS world.
This should not impact the amount of importance given to the proposal, but,
is relevant to the technical debate.
Ravi
--
