dotgnu-auth
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Auth]Identity Services Definition


From: David Sugar
Subject: Re: [Auth]Identity Services Definition
Date: Wed, 02 Jan 2002 04:35:46 -0500
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.4) Gecko/20010914

This looks like a very nice summnation and interesting collection of information. When one looks at single sign-on, I think the most important thing any trustable single signon system must provide is non-corrolative signon. That is, if credentials are presented to site "X" that I am really user "A", these should not be common keyed so that if I go to site "Y", site "X" and "Y" can find out I am the same user and form a common profile of me behind my back even if I have chosen not to provide common identifying/corrolative information.

This is actually not an impossible challege. A good authentication system need only be able to say to site "X" that this user "A" is the same user "A" as before, and say to site "Y" this user "B" is the same user "B", and without needing to provide any common data by using session unique identifiers that "X" and "Y" could then use to determine that "A" and "B" are the same. If the user happens to reveal corrolative information to both such as a common credit card #, etc, well, there are limits as to how much any system can protect the freedom of those that cannot do so for themselves, but the authentication system should not in itself rat out the user at a fundimental level to those that wish to form global profiles of everyone, whether thru the absolute ignorant stupidity of having o0nly one authentication provider or a conceptually poor implimentation that reveals users over multiple providers thru the use of common data during authentication processing.

Adam Theo wrote:

Hi, all. Started work on a "Definition of what makes up a complete Identity system". Requirements, features, frameworks used to create features, etc.

However, I've hit halfway, but lost direction and focus. I started off all organized, with the list: Addressing, Profiles, Authentication, Authorization, Presence, Trust, Discovery. Then I realized that Identity needed to run over a Transport, and often used a Subscription framework to enable Presence and Profile subscriptions, as well as Negotiation to automatically create agreements and find commonalities. I've ended up with a jumbled list of subjects that I'm no longer sure how they relate to each other or in a larger context. After all, Subscription isn't so much a feature as it is a tool to make other features. That's easy, but what about Authorization? I would easily consider that a feature, yet it is used to protect Profiles and Presence information. What about Naming? Is it a feature or a tool to create features?

This is what I have so far: http://www.theoretic.com/identity

I need a new direction and new ideas. I started off thinking it would be an easy list of features, but I'm realizing there is alot more to an Identity system than it's features, and I'm struggling to figure out how to organize it all.

What I first need is a truely complete list of features. Should I make "Single Sign-On" a feature, or keep it as a sub-issue of Authentication? Issues like this need to be resolved, and I need your help. Could everyone here help me? First by making a complete list of features of a "complete" Identity system? By complete I mean "requirements and options", "everything possibly relevant". We can take the next step from there by looking at the relationships between the various features.

Thanks, all. I hope to turn this into a community project that will help DotGNU Virtual Identities and my own Genio [ http://www.theoretic.com/genio ] project with a clear direction and focus.





reply via email to

[Prev in Thread] Current Thread [Next in Thread]