On Wed, Aug 11, 2010 at 4:09 PM, Alan McKinnon <alan.mckinnon@gmail.com>wrote:

> On Thursday 12 August 2010 00:11:12 Bill Longman wrote:
> > On 08/11/2010 01:30 PM, Alan McKinnon wrote:
> > > I refuse to implement password expiration policies and have a vast
> array
> > > of literature to back me up when some dimwit damager gets on his
> > > expiration high horse.
> > >
> > > My users pick their own passwords - I present a list of 5 from apg and
> > > let them pick one. Accounts do expire if they go unused for 90 days,
> but
> > > not passwords.
> > >
> > > What put me onto this policy? I found Gartner recommending password
> > > expiration. I find the best security possible is always the opposite of
> > > what Gartner says. Discovering how the AD admins in the company go
> about
> > > their jobs was the convincing straw :-)
> >
> > The bigger buggerboo I see is the "password complexity" [il]logic.
> > There's this vapid requirement of all these different types of
> > characters needed in one's password, yet the thing you really want to
> > enforce is adequate entropy. If my password is an entire sentence, it
> > will not be brute-forced, even if I used just ASCII A-z. There's just
> > too much key space in 4.7^32. At 10^5 attempts per second, you're likely
> > to find the answer in half a billion years. I hope your keyboard still
> > works, let alone exists....
>
> Your reasoning makes sense, until you consider password length limits
> imposed
> by machines.
>
> Cisco routers authenticating via Tacacs for instance often support nothing
> more than DES hashing <yuck>. The hash routines accept up to 10 characters
> for
> a password but only use the first 8 to calculate the hash.
>
> There are Solaris version nowhere near EOL yet that have similar limits.
>
> All this makes my life as a system integrator cum authenticate go-to guy
> very
> tricky indeed. Luckily management tends to say "Just do what Alan says. It
> makes him shut up and go away".
>
> :-)
>
> p.s. dig the use of "vapid". Wonderful word, truly splendid. Communicates
> in 5
> letters something that takes paragraphs any other way. I shall make a note
> for
> future use.
>
> --
> alan dot mckinnon at gmail dot com
>
> Absolutely. If you do not change your ENCRYPT_METHOD or your PASS_MAX_LEN
in your login.defs file and are still relying on the back end's ability to
safely store your passwords in DES format, well, you're in trouble.