RE: : Issue: differences betweern 3GPP SIP usage and Diam eter SIP document

> - In the PPR in the draft User-Data is mandatory, in the Cx 
> it is optional.
> 
> The only semantics associated to the PPR command is to push a new 
> version of the User Profile contained in the SIP-User-Data 
> AVP. Thus, I 
> understand that this AVP is mandatory, otherwise you either 
> don't send 
> the PPR or you are sending it for other reasons that have no 
> semantics 
> associated. It would be good to understand the motivation behind this.

Motivation is that if only the charging addresses included in 
SIP-Accounting-Information change, then there will be no need to send
the profile in the SIP-User-Data.

> may want to use other mechanisms in the future. I don't 
> understand the 
> reason why Cx makes this optional.

I don't understand that either.  Personally, at this moment, I think
it is an error in 29.229 that should be fixed.

I think it would be good to make sure that the differences are justified
before producing two different protocols for the same purpose.

Regards,

German.

> -----Original Message-----
> From: Miguel Garcia [mailto:[email protected]]
> Sent: jueves, 27 de octubre de 2005 12:22
> To: German Blanco (E2/EEM)
> Cc: Jari Arkko; [email protected]
> Subject: Re: [AAA-WG]: Issue: differences betweern 3GPP SIP usage and
> Diam eter SIP document
> 
> 
> So it would be good to understand why 3GPP has made optional 
> some AVPs, 
> especially when the Diameter SIP application was designed upon the Cx 
> interface in 3GPP. In other words, 3GPP at some point in time has 
> changed, and I don't know why.
> 
> I just picked up a couple of examples:
> 
> - In the PPR in the draft User-Data is mandatory, in the Cx 
> it is optional.
> 
> The only semantics associated to the PPR command is to push a new 
> version of the User Profile contained in the SIP-User-Data 
> AVP. Thus, I 
> understand that this AVP is mandatory, otherwise you either 
> don't send 
> the PPR or you are sending it for other reasons that have no 
> semantics 
> associated. It would be good to understand the motivation behind this.
> 
> - In SIP-Auth-Data-Item SIP-Authentication-Scheme is 
> mandatory, in the 
> Cx it is optional.
> 
> Yes, this is done on purpose. We always need to identify the 
> authentication scheme, becuase that authentication scheme 
> determines the 
> AVPs that will may be included in the SIP-Auth-Data-Item AVP. At the 
> moment only Digest is supported as an aunthentication 
> mechanism, but we 
> may want to use other mechanisms in the future. I don't 
> understand the 
> reason why Cx makes this optional.
> 
> Shall I continue? I think for each of these differences we need to 
> understand that motivation behind 3GPP for the action.
> 
> Regards,
> 
>          Miguel
> 
> 
> 
> German Blanco (E2/EEM) wrote:
> 
> > Miguel,
> > 
> > note that in most of the cases, the AVPs that are listed 
> are optional
> > in Cx and mandatory in the draft, and not the other way 
> around (which
> > would be no problem).  That means the draft is in some way more 
> > restrictive than the Cx.
> > 
> > And the intention in Cx is to allow other authentication 
> schemas, or 
> > at least to be based on a protocol that allows other authentication 
> > schemas, that is why the SIP-Authentication-Scheme is there.
> > 
> > German.
> > 
> > -----Original Message-----
> > From: Miguel Garcia [mailto:[email protected]]
> > Sent: miércoles, 26 de octubre de 2005 17:22
> > To: Jari Arkko
> > Cc: [email protected]; German Blanco (E2/EEM)
> > Subject: Re: [AAA-WG]: Issue: differences betweern 3GPP SIP 
> usage and
> > Diameter SIP document
> > 
> > 
> > Do you want me to reply one by one to all these differences?
> > 
> > Most of the are related to "this AVP is mandatory here and optional 
> > there", and the reason is because 3GPP Cx assumes that 
> there is only 
> > IMS-AKA authentication, where the UE sends an unsolicited 
> Authorization 
> > header in the first SIP request. This is not the case for a 
> regular SIP 
> > client in SIP, which works like in HTTP. So in most cases 
> there is not 
> > going to be an unsoliciated Authorization header, there is 
> not going to 
> > be a realm, username, etc...
> > 
> > I think I don't have the time to comment one by one all the 
> issues. The 
> > only think I can say is that there was a reason why we 
> relaxed the Cx 
> > mandatoriness of AVPs, and there are reasons why we added 
> new AVPs, and 
> > most of these reasons come from moving a Diameter 
> application that works 
> > in a controlled environment (such an IMS) to support the 
> public Internet
> > 
> > /Miguel
> > 
> > Jari Arkko wrote:
> > 
> > 
> >>I know I'm late with this wrt WGLC, but my colleague made a list
> >>of differences between what's in the Diameter SIP document
> >>and what 3GPP Cx interface does.
> >>
> >>Please see the list below. I realize that a standard spec
> >>may not match what someone else has been doing before
> >>and there's bound to be extensions, things done in a better
> >>way, etc.
> >>
> >>However, I'd like circulate this list just to make sure that
> >>we understand what these differences are, and to make
> >>sure that these are intended differences and not accidental
> >>differences that will make it harder for 3GPP to switch to
> >>a IETF standard later (if they even want to, but that's
> >>another discussion).
> >>
> >>Bigger differences:
> >>
> >>- DIAMETER_ERROR_USER_UNKNOWN is used when the User-Name is not
> >>known in the draft, but in Cx when either the User-Name or the
> >>public identity is not known.
> >>
> >>- In the draft, the Authentication Pending flag is cleared in the
> >>MAR. In the Cx, it is cleared in the SAR.
> >>
> >>- In the draft, the SIP Server is stored in the MAR in some cases.
> >>In the Cx, it is only stored in the SAR.
> >>
> >>- In the PPR in the draft User-Data is mandatory,
> >>in the Cx it is optional.
> >>
> >>- In SIP-Auth-Data-Item SIP-Authentication-Scheme is mandatory,
> >>in the Cx it is optional.
> >>
> >>- SIP-User-Data is a Grouped with two mandatory AVPs in the draft,
> >>in the Cx it is an OctetString that contains the XML data.
> >>
> >>- In all responses in the draft Result-Code is mandatory,
> >>in the Cx it is optional as an alternative to Experimental-Result
> >>
> >>- In all draft commands there is a mandatory Auth-Application-Id,
> >>in the Cx there is a mandatory Vendor-Specific-Application-Id.
> >>
> >>Minor differences:
> >>
> >>- In UAR and MAR in the draft User-Name is optional,
> >>in the Cx it is mandatory. The draft specifies an error for when
> >>the application requires a User-Name, but Cx uses a different
> >>error.
> >>
> >>- The draft states that it is not possible to return SIP Server and
> >>capabilities in the same UAA. In the Cx it is done, but this may
> >>be changed.
> >>
> >>- There are two additional responses to MAR in the draft
> >>(DIAMETER_SUCCESS_AUTH_SENT_SERVER_NOT_STORED and
> >>DIAMETER_SUCCESS_SERVER_NAME_NOT_STORED)
> >>
> >>- In the draft, it is possible to keep Diameter sessions and use the
> >>STR and ASR.
> >>
> >>- In UAR, Visited-Network-Identifier is optional in the draft and
> >>mandatory in Cx.
> >>
> >>- In RTR, User-Name is optional in the draft and mandatory in Cx.
> >>
> >>- Format of the Charging Addresses is different.
> >>
> >>- It is possible to generate the once in the Diameter client in the
> >>draft.
> >>
> >>- Authorization-Lifetime and Auth-Grace-Period are present 
> in the draft
> >>in MAA, SAA, PPR, LIA and RTA (are they needed in the RTA at all?).
> >>
> >>- In the LIA in the draft, there are Redirect-Host, 
> Redirect-Host-Usage
> >>and Redirect-Max-Cache-Time, that are not in Cx.
> >>
> >>- There is additional functions in Cx that are not in the 
> draft, like
> >>barring or implicit registration.
> >>
> >>- In the handling of the MAR in the draft, it is stated 
> that the Diameter
> >>server keeps the address of the SIP server.
> >>
> >>
> >>
> > 
> > 
> 
> -- 
> Miguel A. Garcia           tel:+358-50-4804586
> sip:[email protected]
> Nokia Research Center      Helsinki, Finland
> 
>