RE: RADIUS attribute space situation and WGs wanting to add attributes

["Nelson, David" <dnelson@enterasys.com>]

> - What is the state of RADIUS attribute space?
>   Perhaps you can answer with one of the following
>   conclusions:
> 
>      (1) Used up, no near-term possibility to rectify
>          the situation. Please don't charter any new
>          RADIUS work.
>      (2) Used up but a solution will appear. Feel free
>          to recharter, but don't put a milestone until
>          MM, YYYY.
>      (3) Go ahead, no problem.


The RADEXT WG has a charter milestone to address this issue:

   Oct 2006    RADIUS Extended Attributes submitted as a 
   Proposed Standard RFC (split out from Design Guidelines draft)

The WG has recently come to consensus on how to approach this problem.
We are waiting for a formal ID to be circulated.

I think, therefore, that the answer is (b).  The October date might be
achieved with some concerted effort.  A later milestone date in MIP6
would be advisable.   

> - The prospect of providing two sets of attributes instead
>   of one is troubling, particularly if one set of attributes
>   could potentially suffice if those are allocated from
>   RADIUS space.

If there is a real-world, demonstrated need for this feature in existing
RADIUS deployments, it would make sense to have a singe set of
attributes.

There was a very rough consensus of the room at the RADEXT session
during the July IETF meeting that we should not continue to solve new
problems in RADIUS forever.  At some point we would push back on legacy
support in RADIUS and suggest that new applications migrate to Diameter
(only).  While we did not obtain consensus on the WG list, I think we
need some discussion on this issue.