RE: [xacml] Proposed semantics for operations involving INDETERMI NATE

From
Daniel Engovatov <>
Date
2002-07-24T17:27:44+00:00
ID
Thread
RE: [xacml] Proposed semantics for operations involving INDETERMI NATE
Title: RE: [xacml] Proposed semantics for operations involving INDETERMI NATE

They may have it presented the same way for the user - but in this case the equivalent of a PEP is the shell, I think. Is not it?  Logging facilities and the shell do have this information - then they have a choice how to present it..  That's what I thought - PEP has a choice how to present information to the end client, but it still knows the distinction of what happened in PDP.  At least, I would guess that even in this systems, if you examine the system logs you will find different records.

I was talking about having protocol available between PDP and PEP as it

does have use for several variations of recombination algorithms and for

PDP clustering.  How it is presented to the end client - is not it a choice done by an application?   They may choose to behave the same for GRANT and DENY as well - this is out of the scope of the XACML protocol..

-----Original Message-----

From: Anne Anderson - Sun Microsystems

To: xacml

Sent: 7/24/02 7:55 AM

Subject: RE: [xacml] Proposed semantics for operations involving INDETERMI NATE

"Daniel Engovatov" <> wrote:

>Completely disagree.  Every single security system differentiates

between,

>say, "incorrect password" and "service not available".

OSF DCE, Apollo Domain, and HP-UX all are very careful NOT to make any

distinctions (I've worked on lots of other systems, but not recently

enough to speak with reliance on my memory).  The systems I've worked

with

even introduce intentional delays so that a user

can't tell whether a login attempt failed due to invalid user name,

invalid password, some service not available, etc.  This is to avoid

leaking any information that might help an attacker.  Login either

succeeds or it fails, and the time it takes to get a response is the

same

in either case.

Anne Anderson       

Internet Security Research Group, Sun Labs

Sun Microsystems, Inc., Burlington, MA

----------------------------------------------------------------

To subscribe or unsubscribe from this elist use the subscription

manager: <http://lists.oasis-open.org/ob/adm.pl>