Re: Breakin attempts

On Thu, 2010-04-22 at 23:21 -0700, Wolfgang S. Rupprecht wrote:
David Liguori <liguorid@xxxxxxxxxx> writes:
Wolfgang S. Rupprecht wrote:
The core problem is to prevent someone from guessing users' passwords.
You aren't going to achieve real security by hiding this or that
attribute. If you don't want to worry about your users chosing bad
non-random passwords, don't let them. Force them to use a 1k-2k RSA key
for ssh and turn off all login types in sshd_config other than RSA2.
That way any attacker has to correctly guess a 1k-bit computer generated
number. That will almost certainly be much more secure than any
password users will chose. Then you can look at the ssh log files and
laugh. The universe isn't going to last long enough for them to guess
even a small fraction of the keys.

Unless someone builds a quantum computer that can implement the Shor
algorithm for nontrivial cases :-)


I had to look that up. Luckily there are going to be lots of papers
about it if folks can start factoring RSA keys of that length.

More to the point, there would be widespread panic among banks and
online shopping sites, webmail sites, and anywhere else that relies on a
public-key based security model, which is essentially all of them.

Luckily the chances of this happening in the short to medium term seem
very low. IIRC the current record for quantum computers is factoring the
number 15. Getting up to the hundreds of bits is going to be very very
difficult (you can't just string a bunch of smaller ones together like a
conventional computer).


users mailing list
To unsubscribe or change subscription options: