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

FW: Question on Auth-Application-Id AVP on Gx



Title: Question on Auth-Application-Id AVP on Gx
First, my heart-surgery went quite well.
I am back home and occasionally I am scanning email.
I won't react on erveyithing yet... need more time to recover.
 
So, not in good enough shape yet to do detailed checks, I am just forwarding this
to AAA_doctors.
 
Does the below sound acceptable or does it raise alarms?
If it raises alarms, can someone raise via Stephen Hayes, our 3GPP liaison to IETF or
Thomas Narten, our liasison from IETF to 3GPP?
 
Bert
-----Original Message-----
From: Javier Gonzalez Gallego [mailto:ggfj@NORTEL.COM]
Sent: Thursday, February 24, 2005 18:47
To: 3GPP_TSG_CN_WG3@LIST.ETSI.ORG
Subject: Question on Auth-Application-Id AVP on Gx

Dear CN3 delegates
I hope you all have a safe trip back.
I am revising the lates stats of Gx an came across with the following question for the Gx people. Although if my concerns are correct we may experience the same problem across all of the others diameter interfaces we own, that's why I am bringing this to your attention right away so we can decide if there is a problem or not that need to be corrected.

The background for this is that Gx was intended to be DCC Application as defined inIETF, i.e. re-use the DCCA application id. If we go and look in the IETF draft for DCCA, we see that in the CCR/CCA for the exchange of capabilities only the Auth-Application-Id AVP appears and not the Vendor-Specific-Application-Id AVP as defined in Diameter BASE (grouped AVP containing vendor-id and Auth-Application-Id ).  And this is the way it is actually done in Gx: aligned with DCCA

However having a closer look at this, the reason why this is so in IETF is because they only need to publish their own new applications without referring to vendor-id (they don't have vendor id, other 3rd parties such as 3gGPP). So after 3GPP finally decided that Gx was different enough to have its own application-id, we should have probably introduced the vendor-specific-application-id, so both the vendor-id (10415) and the Gx application id are mentioned together under this grouped AVP.

In fact in clause 6 it says

"Gx Messages are carried within the Diameter Application(s) described in the sub-clauses below. These Applications are defined as vendor specific Diameter applications, where the vendor is 3GPP. The vendor identifier assigned by IANA to

3GPP (http://www.iana.org/assignments/enterprise-numbers) is 10415.
In Section 6 of TS 29.210, it states "The TFP and the CRF shall advertise the support of the 3GPP vendor specific Diameter Application for the Gx Application and/or the Gx over Gy Application by including the value of the appropriate application identifier(s) in the Capabilities-Exchange-Request and Capabilities-Exchange-Answer commands."


We don't have placeholders (i.e. AVPs) for the vendor-id=10415.


My feeling at this point is that we should:
A) remove the Auth-Application-Id AVP from command level in CCR and others commands
B) introduce the Vendor-Specific-Application-Id AVP in CCR and other commands, which wil contain the vendo-id and the Auth-Application-Id AVP as definde in RFC 3588


Sorry for the long explanation but I think this is important. Please think about this in case we need to introduce corrections to our specs


   Javier Gonzalez
      UMTS Product Planning
      Nortel Networks
      Tel  +44.(0)1628 434123 (ESN 560 4123)
      >> new domain: ggfj@nortel.com