Re: [OAUTH-WG] JWT Token on-behalf of Use case

[John Bradley <[e-mail]>]

I keep providing links to MS documentation that leads one to a different conclusion for WIF.
https://msdn.microsoft.com/en-us/library/ee748487.aspx <https://msdn.microsoft.com/en-us/library/ee748487.aspx> and https://msdn.microsoft.com/en-us/library/ee517269.aspx
An ActAs RST element indicates that the requestor wants a token that contains claims about two distinct entities: the requestor, and an external entity represented by the token in the ActAs element.

An OnBehalfOf RST element indicates that the requestor wants a token that contains claims only about one entity: the external entity represented by the token in the OnBehalfOf element.

That reflects my understanding that prior to WS-Trust 1.4 OnBehalfOf was impersonation and in WS-Trust 1.4 was added to provide the composite assertion.

The JBOS docs and examples are also consistent with this https://docs.jboss.org/author/display/JBWS/ActAs+WS-Trust+Scenario
Or these: https://thilinamb.wordpress.com/2009/08/21/identity-delegation-in-ws-trust-1-4/
                http://blogs.msdn.com/b/alikl/archive/2011/03/15/windows-identity-foundation-wif-the-difference-between-actas-and-onbehalfof.aspx?Redirected=true
                https://www.netiq.com/documentation/netiqaccessmanager4/identityserverhelp/data/ws-trust_usecases.html

Given that you were one of the authors, you are as close to authoritative as one could get.   

However as far as I can tell there is some confusion that needs to be resolved. 

I agree we need the functionality of the two, and I don’t really care what they are called as long as we are not confusing developers.


John B.

> On Jul 6, 2015, at 5:43 PM, Anthony Nadalin <[e-mail]> wrote:
> 
> The WS-Trust “ActAs” mimics the Windows Kerberos Protocol Transition (impersonation)  feature as this enables an account to impersonate another account for the purpose of providing access to resources. In a typical scenario, the impersonating account would be a service account assigned to a web application or the computer account of a web server. The impersonated account would be a user account requiring access to resources (e.g. data in an SQL database) via a web application. In this scenario, SQL server would be accessed by the impersonating (service account) account, however access would be under the context of the impersonated (user) account.
>  
> WS-Trust “OnBehalfOf”  mimics the Windows Kerberos Constrained Delegation feature, which lets you limit the back-end services for which a front-end service can request tickets on behalf of another user. “OnBehalfOf”  allows a selected services on a server can be granted for access by the impersonating account, whilst other services on the same server, or services on other servers are denied for access.
>  
> Maybe someone can summarize why they think the text for ActAs and OnBehalfOf in WS-Trust or Windows Kerberos is wrong or swapped as I have not seen a clear explanation other than John saying that Brian knows and Brian saying John knows.  <>
>  
> Our usage and use cases are based upon the deployed services of WS-Trust and Kerberos support in Windows (workstation and server) and Xbox.
>  
> From: OAuth [mailto:[e-mail]] On Behalf Of Brian Campbell
> Sent: Monday, July 6, 2015 11:29 AM
> To: Mike Jones <[e-mail]>
> Cc: oauth <[e-mail]>
> Subject: Re: [OAUTH-WG] JWT Token on-behalf of Use case
>  
> Stating specific action items resulting from the ad-hoc meeting in Dallas like that suggests some clear consensus was reached, which is not at all the case. As I recall, several of us argued past one another for an hour or so and decided to adjourn in order to go to the bar (okay, and dinner too - but mostly beer).
> 
> The impression about reversal of terms, I think, comes from the text in https://tools.ietf.org/html/draft-ietf-oauth-token-exchange-01#section-1.3 <https://tools.ietf.org/html/draft-ietf-oauth-token-exchange-01#section-1.3> which hurts my head a little every-time I read it but does seems to confuse things. The MSDN link <https://msdn.microsoft.com/en-us/library/ee748487.aspx> John gave is much more to the point than WS-Trust (I don't believe WS-Trust can be pointed to as a model of clarity).  In the draft I wrote, I tried to take Mike's text and clarify a distinction between impersonation and delegation with https://tools.ietf.org/html/draft-campbell-oauth-sts-01#section-1.3 <https://tools.ietf.org/html/draft-campbell-oauth-sts-01#section-1.3> and then also be very explicit about act-as vs. on-behalf-of in the parameter definitions at https://tools.ietf.org/html/draft-campbell-oauth-sts-01#section-2 <https://tools.ietf.org/html/draft-campbell-oauth-sts-01#section-2> in a manor that was consistent with WS-Trust and the MSDN explanation. I could see value in breaking with that shaky legacy and using new terms too. But I get the point of trying to keep with the old also and potential for even more confusing by using new terms. 
> 
> I wrote draft-campbell-oauth-sts last year in response to the call for adoption of jones-oauth-token-exchange (thread from the archive <https://www.ietf.org/mail-archive/web/oauth/current/msg13305.html>). Though I didn't try and stand in the way and indicated a willingness to collaborate on things. With the expectation, of course, that the details would differ from the -00s and -01s as work progressed. Folks seemed generally amenable to that <https://www.ietf.org/mail-archive/web/oauth/current/msg13308.html> at the time but little has happened since then.
> 
> Phil's earlier point about the priory of this getting pushed down has some truth to it. But I still believe it's something that can provide a lot of value in standardizing, if we do so in a reasonable way.
>  
> 
> 
> 
> 
>  
> 
>  
> On Mon, Jul 6, 2015 at 10:33 AM, Mike Jones <[e-mail] <mailto:[e-mail]>> wrote:
> It would surprise me if on-behalf-of and act-as were reversed with respect WS-Trust, because the explanations of the terms came directly from WS-Trust 1.4.  I also think the chances of us reducing confusion by inventing new terminology, rather than adding to it, would not be in our favor. :-/
> 
> FYI, the action items outstanding from our ad-hoc meeting on this draft in Dallas are:
>   - Allowing security types other than JWT to also be used as the act_as and on_behalf_of request values.
>   - Further integrating the mechanism into the existing OAuth ecosystem - allowing use of access tokens or refresh tokens when appropriate.
> 
> I plan to do the first today.  The second is probably more than I'll get done today before the submission cutoff.  I agree with John that it would be useful to have discussions on this in Prague on the best way to achieve this further integration.  I'll plan to come into the Prague meeting with a concrete proposal for review.
> 
>                                 Best wishes,
>                                 -- Mike
> 
> -----Original Message-----
> From: OAuth [mailto:[e-mail] <mailto:[e-mail]>] On Behalf Of John Bradley
> Sent: Monday, July 06, 2015 8:13 AM
> To: Brian Campbell
> Cc: oauth
> Subject: Re: [OAUTH-WG] JWT Token on-behalf of Use case
> 
> Yes unfortunately we haven’t made any progress on this since accepting Mike’s first draft.
> 
> His proposal is basically for a new endpoint while Brian tired to fit it into the existing token endpoint.
> 
> I think draft-ietf-oauth-token-exchange-01 still has OnBehalfOf and ActAs reversed compared to WS-Trust 1.4.
> see https://msdn.microsoft.com/en-us/library/ee748487.aspx <https://msdn.microsoft.com/en-us/library/ee748487.aspx> for the short explanation.
> 
> I think Brian is closer in explaining it.
> 
> In fairness because WS-Trust originally only had On-Behalf-Of the naming and what people put in tokens is a bit muddled in many implementations.
> I think many times it is how WIF implemented it that people copied.
> 
> It may be better to have new terms that are clear such as impersonation and composite.
> 
> The WG needs to decide if this is going to be an entirely new endpoint, free of the Token endpoint semantics.   There are plusses and minuses to both options.
> 
> Also while it is nice to be pure and talk about abstract security tokens, it would be good to give some guidance on what a composite security token would look like for interoperability.
> 
> There are also issues around how this would work with proof of possession security tokens, both as input and output.
> 
> Perhaps we can make some progress on this in Prague.
> 
> John B.
> 
> 
> 
> 
> > On Jul 6, 2015, at 11:04 AM, Brian Campbell <[e-mail] <mailto:[e-mail]>> wrote:
> >
> > Thanks Sergey,
> >
> > The goal of draft-campbell-oauth-sts was to be consistent with OAuth 2.0 and thus hopefully familiar to developers and easy to understand and implement (especially from the client side). It's also intended to be flexible in order to accommodate a variety of use-cases including the chaining type cases that Justin's draft covers.
> >
> > Specifying a security_token_type of the returned token is just a way of providing more info to the client about the token (i.e. is this a JWT or a SAML token or something else) via a URI. It's not always needed but in STS style cases the tokens are not always opaque to the client and the parameter just provides info about the returned token.
> >
> > On Mon, Jul 6, 2015 at 5:33 AM, Sergey Beryozkin <[e-mail] <mailto:[e-mail]>> wrote:
> > Hi Brian
> >
> > I've read the text, I like it is still pure OAuth2, with few extra parameters added to the access token request, and a key response property being 'access_token' as opposed to 'security_access_token' as in the draft-ietf-oauth-token-exchange-01.
> > It appears draft-campbell-oauth-sts-01 can cover a draft-richer-oauth-chain-00 case with the on_behalf_of (and/or act_as ?) property being an original client token but not 100% sure given draft-richer-oauth-chain-00 covers a specific case.
> >
> > One thing I'm not sure about is what is the purpose of specifying a
> > security_token_type of the returned access token
> >
> > Thanks, Sergey
> >
> > On 01/07/15 15:59, Brian Campbell wrote:
> > One problem, I think, with token exchange is that it can be really
> > simple (token in and token out) and really complicated (client X wants
> > a token that says user A is doing something on behalf of user B) at
> > the same time.
> >
> > I put forth https://tools.ietf.org/html/draft-campbell-oauth-sts-01 <https://tools.ietf.org/html/draft-campbell-oauth-sts-01> in
> > an attempt to simplify things and express what I envisioned as an
> > OAuth based token exchange framework. Though it likely only muddied
> > the waters :)
> >
> > On Wed, Jul 1, 2015 at 7:07 AM, Sergey Beryozkin <[e-mail] <mailto:[e-mail]>
> > <mailto:[e-mail] <mailto:[e-mail]>>> wrote:
> >
> >     Hi Justin
> >
> >     https://tools.ietf.org/html/draft-richer-oauth-chain-00 <https://tools.ietf.org/html/draft-richer-oauth-chain-00> is much
> >     easier to read, that I can tell for sure, at least it is obvious why
> >     a given entity (RS1) may want to exchange the current token provided
> >     by a client for a new token. Definitely easily implementable...
> >
> >     One thing I'm not sure in the draft-richer-oauth-chain-00 about is
> >     on behalf of whose entity RS1 will be acting once it starts
> >     accessing RS2, On Behalf Of RO, or may be On Behalf Of (RO +
> >     Client), or may be it is On Behalf Of RO + Act As Client ? The last
> >     one seems most logical to me...
> >
> >     Thanks, Sergey
> >
> >
> >     On 01/07/15 13:18, Justin Richer wrote:
> >
> >         As it's written right now, it's a translation of some WS-*
> >         concepts into
> >         JWT format. It's not really OAuth-y (since the client has to
> >         understand
> >         the token format along with everyone else, and according to the
> >         authors
> >         the artifacts might not even be "OAuth tokens"), and that's my main
> >         issue with the document. Years ago, I proposed an OAuth-based
> >         token swap
> >         mechanism:
> >
> >         https://tools.ietf.org/html/draft-richer-oauth-chain-00 <https://tools.ietf.org/html/draft-richer-oauth-chain-00>
> >
> >         This works without defining semantics of the tokens themselves, just
> >         like the rest of OAuth. I've proposed to the authors of the current
> >         draft that it should incorporate both semantic (using JWT) and
> >         syntactic
> >         (using a simple token-agnostic grant) token swap mechanisms, and
> >         that
> >         the two could be easily compatible.
> >
> >            -- Justin
> >
> >         On 7/1/2015 6:35 AM, Sergey Beryozkin wrote:
> >
> >             Hmm... perhaps the clue is in the draft title,
> >             token-exchange, so may
> >             be it is a case of the given access token ("on_behalf_of" or
> >             "act_as"
> >             claim) being used to request a new security token. One can
> >             only guess
> >             though, does not seem like the authors are keen to answer
> >             the newbie
> >             questions...
> >
> >             Cheers, Sergey
> >
> >
> >             On 30/06/15 13:38, Sergey Beryozkin wrote:
> >
> >                 Hi,
> >                 Can you please explain what is the difference between
> >                 On-Behalf-Of
> >                 semantics described in the
> >                 draft-ietf-oauth-token-exchange-01 and the
> >                 implicit On-Behalf-Of semantics a client OAuth2 token
> >                 possesses ?
> >
> >                 For example, draft-ietf-oauth-token-exchange-01 mentions:
> >
> >                 "Whereas, with on-behalf-of semantics, principal A still
> >                 has its own
> >                 identity separate from B and it is explicitly understood
> >                 that while B
> >                 may have delegated its rights to A, any actions taken
> >                 are being taken by
> >                 A and not B. In a sense, A is an agent for B."
> >
> >                 This is a typical case with the authorization code flow
> >                 where a client
> >                 application acts on-behalf-of the user who authorized
> >                 this application ?
> >
> >                 Sorry if I'm missing something
> >
> >                 Cheers, Sergey
> >                 On 25/06/15 22:28, Mike Jones wrote:
> >
> >                     That’s what
> >                     https://tools.ietf.org/html/draft-ietf-oauth-token-exchange-01 <https://tools.ietf.org/html/draft-ietf-oauth-token-exchange-01>
> >                     is
> >                     about.
> >
> >                     Cheers,
> >
> >                     -- Mike
> >
> >                     *From:*OAuth [mailto:[e-mail] <mailto:[e-mail]>
> >                     <mailto:[e-mail] <mailto:[e-mail]>>] *On Behalf Of *Vivek
> >                     Biswas
> >                     -T (vibiswas - XORIANT CORPORATION at Cisco)
> >                     *Sent:* Thursday, June 25, 2015 2:20 PM
> >                     *To:* [e-mail] <mailto:[e-mail]> <mailto:[e-mail] <mailto:[e-mail]>>
> >                     *Subject:* [OAUTH-WG] JWT Token on-behalf of Use
> > case
> >
> >                     Hi All,
> >
> >                         I am looking to solve a use-case similar to
> >                     WS-Security On-Behalf-Of
> >
> > <http://docs.oasis-open.org/ws-sx/ws-trust/v1.4/errata01/os/ws-trust-1 <http://docs.oasis-open.org/ws-sx/ws-trust/v1.4/errata01/os/ws-trust-1>
> > .4-errata01-os-complete.html#_Toc325658980>
> >
> >
> >                     with OAuth JWT Token.
> >
> >                         Is there a standard claim which we can define
> >                     within the OAuth JWT
> >                     which denote the On-behalf-of User.
> >
> >                     For e.g., a Customer Representative trying to create
> >                     token on behalf of
> >                     a customer and trying to execute services specific
> >                     for that specific
> >                     customer.
> >
> >                     Regards,
> >
> >                     Vivek Biswas,
> >                     CISSP
> >
> >                     *Cisco Systems, Inc <http://www.cisco.com/ <http://www.cisco.com/>>*
> >
> >                     *Bldg. J, San Jose, USA,*
> >
> >                     *Phone: [phone] <tel:%2B1%20408%20527%209176>
> > <tel:%2B1%20408%20527%209176 <tel:%2B1%20408%20527%209176>>*
> >
> >
> >
> >                     _______________________________________________
> >                     OAuth mailing list
> >                     [e-mail] <mailto:[e-mail]> <mailto:[e-mail] <mailto:[e-mail]>>
> >                     https://www.ietf.org/mailman/listinfo/oauth <https://www.ietf.org/mailman/listinfo/oauth>
> >
> >
> >
> >             _______________________________________________
> >             OAuth mailing list
> >             [e-mail] <mailto:[e-mail]> <mailto:[e-mail] <mailto:[e-mail]>>
> >             https://www.ietf.org/mailman/listinfo/oauth <https://www.ietf.org/mailman/listinfo/oauth>
> >
> >
> >         _______________________________________________
> >         OAuth mailing list
> >         [e-mail] <mailto:[e-mail]> <mailto:[e-mail] <mailto:[e-mail]>>
> >         https://www.ietf.org/mailman/listinfo/oauth <https://www.ietf.org/mailman/listinfo/oauth>
> >
> >
> >     _______________________________________________
> >     OAuth mailing list
> >     [e-mail] <mailto:[e-mail]> <mailto:[e-mail] <mailto:[e-mail]>>
> >     https://www.ietf.org/mailman/listinfo/oauth <https://www.ietf.org/mailman/listinfo/oauth>
> >
> >
> >
> >
> >
> > _______________________________________________
> > OAuth mailing list
> > [e-mail] <mailto:[e-mail]>
> > https://www.ietf.org/mailman/listinfo/oauth <https://www.ietf.org/mailman/listinfo/oauth>
> 
> _______________________________________________
> OAuth mailing list
> [e-mail] <mailto:[e-mail]>
> https://www.ietf.org/mailman/listinfo/oauth <https://www.ietf.org/mailman/listinfo/oauth>
>  
> _______________________________________________
> OAuth mailing list
> [e-mail] <mailto:[e-mail]>
> https://www.ietf.org/mailman/listinfo/oauth <https://www.ietf.org/mailman/listinfo/oauth>