Kerberos Feature Request

[Henry B. Hotz]

Multiple issues here.  All good points, but one at a time:

1)  I am not requesting that anybody support Microsoft PAC for 
authorization.  I am only requesting a general interface for 
inserting authorization data.  As noted by Paul W. Nelson a complete 
solution also needs a good API for a client service to pull that 
authorization data out and inspect it (including parsing routines, 
etc.).

2)  I am requesting that the general interface be usable to 
distribute Microsoft PAC information, but only if other out-of-scope 
work is done to ensure that data is compatible with what Microsoft 
wants.

3)  The format of the PAC data is claimed to be proprietary and trade 
secret by Microsoft, and is subject to unilateral change by 
Microsoft.  However Microsoft has built an infrastructure based on 
the PAC data.  They aren't going to break backward compatibility if 
they don't have to.

4)  I note that organizations or people who have agreed to 
Microsoft's license on their description of the PAC data may have to 
abide by their agreements.  However I've discussed the issue with 
JPL's legal department and I'm satisfied that *I* do not have to 
worry about that.  The PAC format is readily available from multiple 
sources (see Google) and you don't have to agree to anything in order 
to get the information.  Some fraction (I won't say how much) of 
Microsoft's legal claims on that information are based on the 
information being a "trade secret".  It's no longer a secret, trade 
or otherwise.

Usual disclaimer:  I'm not a lawyer, I don't even play one on TV (or 
in the UK ;-).

At 8:53 AM +0000 2/11/04, Tim Alsop wrote:
>Henry,
>
>The PAC format is a Microsoft propriatory standard that they have 
>documented for vendors to use if they wish. If Microsoft decide to 
>change this, they can - they will then inform all MSDN members that 
>changes have been made.
>
>To implement your proposal - each KDC would have to construct that 
>PAC field and place this into the appropriate tickets. The KDC would 
>also have to manage principals such as ldap/host.name at REALM and 
>cifs/server.name at REALM. The additional service principals would need 
>to be created when a workstation becomes a member of a domain etc. I 
>am sure there are many other complex implications here. Effectively 
>what you are asking for is a non-MS kdc that is usable instead of MS 
>AD for workstation user logon ?
>
>One big issue, IMHO - What do you think Microsoft would think if it 
>was possible to use a UNIX server with an LDAP server instead of 
>Active Directory ? They would clearly not like this and they might 
>decide to change licensing of PAC data (or some other action) to 
>stop the UNIX KDC vendors from being able to offer this 
>functionality ... Do you agree ?
>
>Cheers, Tim.
>
>-----Original Message-----
>From: Henry B. Hotz [<mailto:hotz at jpl.nasa.gov>mailto:hotz at jpl.nasa.gov]
>Sent: 11 February 2004 01:29
>To: Tim Alsop; krbdev at mit.edu; heimdal-discuss at sics.se; 
>darwin-development at lists.apple.com
>Cc: Dj Byrne
>Subject: RE: Kerberos Feature Request
>
>I want to enable that.
>
>I'm suggesting that it would be nice if there were a MIT-independent 
>and KTH-independent (and CyberSafe-independent ;-) mechanism that 
>allowed you to do that.  Given a KDC-neutral enabling mechanism I 
>expect that an open-source project or 10 would spontaneously form to 
>bridge the gap between the conformant KDCs and the LDAP server of 
>your choice (including true blue AD).
>
>I'd be happy if the agreement/standard/whatever just said that you 
>do an ldap query for the "pac" attribute from the unique ID that 
>matches the principal, with the obvious REALM to DC= translation.
>
>Jeffrey Altman objects that I want an API, not an RFC, so IETF 
>shouldn't be involved, but I think the example I just gave would be 
>an RFC.  I'm trying to limit my care-about's though.  I just want a 
>general way to make use of the feature, which is currently pretty 
>inaccessible.
>
>At 11:41 PM +0000 2/10/04, Tim Alsop wrote:
>>Henry,
>>
>>Are you proposing that the non-Microsoft KDC issues tickets containing
>>PAC data and gets the group membership information from the Active
>>Directory using LDAP ?
>>
>>Thanks, Tim.
>>
>>-----Original Message-----
>>From: Henry B. Hotz
>>[<<mailto:hotz at jpl.nasa.gov>mailto:hotz at jpl.nasa.gov><mailto:hotz at jpl.nasa.gov>mailto:hotz at jpl.nasa.gov]
>>Sent: 10 February 2004 18:27
>>To: krbdev at mit.edu; Tim Alsop; heimdal-discuss at sics.se;
>>darwin-development at lists.apple.com
>>Cc: Dj Byrne
>>Subject: Kerberos Feature Request
>>
>>I probably should send this to the IETF group, but I'm not on their
>>mailing lists.  (Apologies if the cross-posting causes problems.) It
>>would be *nice* if all Kerberos distributions added this feature the
>>same way.
>>
>>One of the famous things that Microsoft did in their AD Kerberos
>>implementation is added authorization data to the (supposedly
>>
>>optional) PAC field that is necessary when using certain other
>>Microsoft functionality.  AFAIK all of the information added is also
>>contained in the LDAP directory that AD also provides.
>>
>>I do not think it makes any sense for a (non-Microsoft) Kerberos server
>>to directly maintain this data.  Rather it should have a mechanism for
>>acquiring the data from an external source, such as an LDAP directory.
>>
>>My request is that the Kerberos community agree on a standard external
>>interface to get that data.  If the interface itself were standardized
>>then the work of connecting that interface to the appropriate AD
>>attributes could be done independently of any Kerberos server, and
>>could be updated as Microsoft updates their schema independent of
>>Kerberos versions.  It would also make the use of PAC data in
>>non-Microsoft environments much easier to consider.
>>
>>--
>>The opinions expressed in this message are mine, not those of Caltech,
>>JPL, NASA, or the US Government.
>>Henry.B.Hotz at jpl.nasa.gov, or hbhotz at oxy.edu
>
>
>--
>The opinions expressed in this message are mine, not those of 
>Caltech, JPL, NASA, or the US Government.
>Henry.B.Hotz at jpl.nasa.gov, or hbhotz at oxy.edu


-- 
The opinions expressed in this message are mine,
not those of Caltech, JPL, NASA, or the US Government.
Henry.B.Hotz at jpl.nasa.gov, or hbhotz at oxy.edu