>= It is not all that difficult for an attacker to arrange to sniff lots
>  of AS exchanges.  Again, I'm not comparing it to `ypcat passwd' ---
>  but one can't just sweep this possibility under the carpet.

One thing that may not be self-evident is that it's a lot harder to do
this in V5 than in V4.

- Each key is salted with the whole principal name, which means you can't
  build up a sequence of trial keys from passwords; you need to generate
  them on a per-user basis.
- Each encrypted sequence has a confounder, which means you need to do
  more trial decryptions before you can figure out if a key is right
  or not.
- V5 includes MUCH better hooks for password quality control (including
  things like maximum and minimum password lifetimes and a password
  history).

So no, this does not eliminate the problem.  But I believe it reduces it
to a manageable level.  But I doubt you're going to find a security
system that provides absolute security (I know Tom Wu would love for us
all to believe that SRP is such a system, but somehow I'm skeptical).

--Ken

Reply via email to