Re: [ldapext] [ldap] Re: draft-stroeder-hashed-userpassword-values-01

[Andrew Findlay <[e-mail]>]

On Thu, Mar 14, 2013 at 08:19:12PM +0100, Michael Ströder wrote:

> >>      userpasswordvalue  = cleartext-password / prefix hashed-password
> > 
> > I think you should replace "hashed-password" with "scheme-specific data" and
> > stop there.
> 
> That's a conclusion of your opinion. But I want to describe the *order* of
> password and salt used by any server I saw using the scheme.

Why not separate the description of the data from the overall syntax?
It will be easier to read that way, and much more obvious that the whole
thing is extensible and a bit informal.

userPassword has Octet String syntax, so in principle the value is
<scheme name in curly brackets> <arbitrary data>

A separate section of the doc could then describe (or refer to) the formats
of all the commonly-used storage schemes. I was about to call them 'hash
schemes' but that is wrong, as some servers support reversible encryption
schemes as well as hashes.


On a slight tangent, a rough guide to the current strength of various hash
schemes can be found on hashcat's front page:

	http://hashcat.net/oclhashcat-plus/

The table at the bottom gives the brute-force attack rate in crypts/sec
using a single PC with a good (mid-range gaming) graphics engine.
Numbers range from about 4k c/s for bcrypt up to 7500M c/2 for NTLM.
It does not explicitly list figures for SSHA and SMD5 but I suspect the
'sha512crypt $6$' figure is indicative at 12k c/s.

Andrew
-- 
-----------------------------------------------------------------------
|                 From Andrew Findlay, Skills 1st Ltd                 |
| Consultant in large-scale systems, networks, and directory services |
|     http://www.skills-1st.co.uk/                [phone]     |
-----------------------------------------------------------------------