> -----Original Message-----
> From: Richard [mailto:[email protected]] 
> Sent: Monday, July 27, 2009 2:57 PM
> To: Dan Harkins
> Cc: Joseph Salowey (jsalowey); [email protected]; 
> [email protected]
> Subject: Re: [Emu] If we use the Radius property extension
> 
> Hi, Dan:
>  
> Thanks for your comments. Please check my comments inline 
> which is identified by [Richard].
> 
> 
> 2009/7/28 Dan Harkins <[email protected]>
> 
> 
> 
>        Hi Richard,
>       
>       [Dan] On slide 11 you say you want to reuse the AAA 
> architecture.
>       But EAP is not a AAA protocol. RADIUS is but you don't want to
>       use RADIUS. So I don't understand what reuse you're hoping to
>       take advantage of.
> 
>    
> //////////////////////////////////////////////////////////////
> ////////////
>    [Richard] Yes, I want to reuse the AAA architecture. This 
> is a main objective
>    and the reason for it is described on the slide 9.  For 
> the reusing,
>    there are two methods. One is the Radius property 
> extension and it is a direct
>    way to reuse the AAA; Another way is to make the WAI over 
> EAP. As the Radius 
>    has a EAP property, so WAI over EAP establish a bridge to 
> reuse the AAA. 
> 
[Joe] EAP is a protocol to authenticate an EAP Peer that communicates
between the EAP Peer and an EAP Server.  EAP is not an alternative to a
AAA protocol.   

>       
>       [Dan]Since EAP doesn't have a transport how do you 
> propose to send
>       EAP packets from the AP to the server? Over UDP port 3810? The
>       same port you send WAI packets to now? If that's the case then
>       all EAP is doing is providing a needless encapsulation of your
>       WAI packet. You gain nothing by encapsulating your WAI packet
>       with an EAP header.
> 
>    
> //////////////////////////////////////////////////////////////
> //////////
> 
>       [ Richard] When using the WAI over EAP, the AP would 
> transport EAP packet 
> 
>       (contains WAI) over Radius. 
> 
>    Also, Please check my last email: It compares the solutions of
>    the Radius Property extension with the WAI over EAP. 
>    The main advantage of WAI over EAP is to keep the loose coupling
>    between the AAA and the WAPI.
>    The reason is:
>   Radius server need not check things going on with WAPI. (EAP-WAI).
>   There is a standard interface  (EAP-SUCCESS or EAP-FAILURE)
>   to let Radius know the result of the authentication.
>   From the AAA server perspective, like the other 
> authentication methods  such as EAP- TLS,  there is no any 
> special handle required for EAP-WAI.
>  
[Joe] EAP-Success and EAP-Failure are not part of RADIUS. They are part
of EAP messages that are for the peer, which is the client being
authenticated.  RADIUS has an Access-Accept and Access-Reject messages
for informing the RADIUS client of its decision.  Your proposal does not
appear to be an authentication mechanism like EAP-TLS since it does not
communicate with the peer.  

Can you give an example of what you mean by loose coupling between AAA
and WAPI?  


>  [Dan] I do have a suggestion to fix this problem. Just get 
> rid of the "certificate authentication" exchange and have the 
> client validate the AP's certificate and the AP validate the 
> client's certificate.
> It is not necessary to have a central authority to approve of 
> certificates, that's the whole point in using certificates.
> ///////////////////////////////////////////////////////////////////
> [ Richard] Yes, your suggestion could work well without a a 
> central authority involved.
> While the current problem I try to resolve is: The WAPI 
> protocol already exist (it needs a central authority to 
> valide the certificate), to support (follow) it and enable it 
> work well under the AAA architecture, I brougt forward the 
> method of WAI over EAP. 
> To avoid the change of the station, and minimize the influnce 
> to the customers, the solution would only change the packet 
> between AP (authenticator) and AAA server.
> To achieve this objective and avoid the influnce to the AAA, 
> the AP should minic the station.
> For the WAPI protocol, please see the slide 7.
> 
[Joe] This is not the way EAP works currently.  EAP is a protocol that
involves the peer.  If you have the authenticator mimic the peer then it
is not EAP in its current form. 


> 
>       
>       
>        regards,
>       
>        Dan.
>       
> 
>       On Mon, July 27, 2009 7:54 am, Richard wrote:
>       > Hi, all:
>       >
>       > The link to the slide:
>       > http://www.ietf.org/proceedings/75/slides/emu-10.ppt
>       > Thanks for your comments in the EMU session.
>       >
>       > Yes, the Radius property extension could be one solution.
>       > The problem to it is:
>       > The Radius has to handle the WAPI packet (Certificate 
> authentication
>       > response),
>       > it would lead to the tight coupling between Radius 
> and WAPI (ASE).
>       >
>       > Compared to it, the EAP solution is simple and make 
> it loose coupling with
>       > WAPI.
>       > Radius server need not check things going on with 
> WAPI. (EAP-WAI).
>       > There is a standard interface to let Radius to know the result
>       > of authentication (EAP-SUCCESS or EAP-FAILURE).
>       > From the AAA server perspective, there is no any 
> special for EAP-WAI
>       >  to the other authentication methods such as EAP-TLS.
>       > The loose coupling for EAP solution is a main 
> advantage to the Radius
>       > extension
>       > method.
>       >
>       > Yes, the EAP framework is to authenticate the station 
> (supplicant). I
>       > guess
>       > the
>       > main reason is that most authentication methods are 
> to authenticate the
>       > supplicant,
>       > and the behavior of Authenticator is pass-through.
>       > Now, WAPI is a use case of authenticating the 
> authenticator (the behavior
>       > of
>       > Authenticator
>       >  is not Pass-through). And it may have more protocols 
> which have similar
>       > behavior
>       > (use cases) like the WAPI.
>       > Could we allow the EAP method behavior like the WAI 
> over EAP instead of
>       > forbidding it?
>       >
>       > Any way, the problem I am trying to resolve is to 
> reuse the AAA and
>       > avoid the tight coupling between Radius and WAPI (ASE).
>       >
>       > Do you have any better solution to it?
>       >
>       > Regards
>       > Richard
>       
>       > _______________________________________________
>       > Emu mailing list
>       > [email protected]
>       > https://www.ietf.org/mailman/listinfo/emu
>       >
>       
>       
>       
> 
> 
> 
_______________________________________________
Emu mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/emu

Reply via email to