RE: iFCP: security position

["Bill Strahm" <bill@sanera.net>]

WEPs problem was not a weakness in encryption security, heck the crypto is
rock solid 128 bit used in every SSL connection on the internet (including
all of your stock transactions, credit card transactions etc.).  Note that
the cryptography is FINE.

What was not fine was the system built around it, specifically, there was no
rekeying algorithm (bad) and they deployed it in such a way that as soon as
you saw a little over a million packets on the wire, it was broken by
default.

The next thing that tends to break crypto systems is random number
generation, there were many hacks on Kerberos based on the usage of a
timestamp to initialize the random number generator.

The third thing that tends to break crypto systems is social engineering
(Please give me your password tends to work about 25% of the time when
random people start calling into your company claiming to be I.T.)

WAY down the list is actually breaking the cipher...  Ok, given 100K and 22
hours, I can break DES... However if my data is only worth 10K and I cange
keys often, then this is acceptable.

Again it is up to the administrator to determine what the acceptable
crytography is.  Heck I use VERY good crypto, but then I have fast machines,
and live in a country that lets me use it.  Until the IPsec WG removes DES
as a MUST implement, I am sorry but it will be in every conforming IPsec
implementation out there.

Bill
Sanera Systems Inc.

-----Original Message-----
From: Paul Koning [mailto:pkoning@jlc.net]
Sent: Monday, September 10, 2001 10:50 AM
To: bill@Sanera.net
Cc: ips@ece.cmu.edu
Subject: RE: iFCP: security position


Excerpt of message (sent 7 September 2001) by Bill Strahm:
> Why do you care how traffic is encrypted ???
>
> Would you rather see Clear traffic than DES traffic ?

Yes, absolutely.

That is because clear traffic does not mislead.  It is
obviously not secure.  DES is sufficiently weak that encrypting with
it could be viewed as a form of false advertising.

This is also what is wrong with things like WEP -- these are systems
that pretend to offer security but in fact do not.  And people defend
them with similar arguments.  Or, for that matter, Fred Foobar's
Famous Snake Oil encryption algorithm.  The problem in all these cases
is that the appearance of crypto without the reality is much, much
worse than the absence of crypto.  You should have either strong
crypto, or none.  After all, strong crypto is readily available.

DES shows up as mandatory in IPsec for reasons that were political, not
technical, and that became obsolete several years ago.

     paul