Re: Comments on draft-zhang-mip6-find-reg-ha-00.txt

Zhang Jian wrote:

>>It seems that even basic RFC 3588 could already do a part of this,
>>too, or at least the AAA server will be aware of what HA is being
>>used.
>>
>>It wasn't clear to me if draft-giaraetta allows a random application server
>>to retrieve node X's current HA via AAA. Does it?
>>
>>    
>>
>I'm not sure whether Diameter can provide this function. Even it can provide
>this function, I think it can only support AAA service. At present, AAA server
>don't aware of what HA is being used, because no specification describes this
>mechanism. 
>  
>
Diameter servers are capable of sending messages to the entity that
is providing the service (NAS or HA or AP or ...). As a part of this,
they have to be aware where the service is being provided.

>Draft-giaraetta only support AAA service. It do not allows a random application
>server to retrieve nods X's current HA vis AAA.
>  
>
I think there are two parts - the infrastructure (= AAA server) finding
out, and the application servering finding out this from the infrastructure.
I do not think the latter has been specified anywhere.

But, as you say, we might not even want to couple this to AAA.

The reason that I brought it up was that there may be a subset
of services where the existing AAA mechanisms are sufficient.
For instance, any location-based logic at the AAA server side
would probably be doable. One of the decisions that we have to
make regarding this work is whether there's sufficient demand
to actually do something in this space, or whether the existing
mechanisms cover the main needs already.

>>>=>The draft-ietf-geopriv-radius-05.txt does not support mobile IPv6.
>>>Am I right? 
>>> 
>>>
>>>      
>>>
>>Depends. The geopriv draft allows a "service node" (e.g. HA, NAS, AP) to
>>tell
>>the AAA network where the client is. It can do this both initially, as well
>>as update the information during the session. Does this suffice for what
>>you want to do? Presumably this depends on the requirements...
>>
>>    
>>
>In the basic specification of mobile IPv6, there is not description about 
>wheter
>it is needed to do authentication and authorization again while mobile node 
>hand over.  If it is needed, it will increase the delay of handover.
>The location service in our draft should be call as location management.
>There may be some other service that need this mechanism in future.
>  
>
Sure. I agree that you shouldn't need to re-authenticate. But that
doesn't rule out the home agent sending some parallel messages
to some infrastructure component.

--Jari