RE: [MIP-QOS] MIP QoS requirements

[Hemant.Chaskar@nokia.com]

Hi Phil,

Some remarks:

> -----Original Message-----
> From: ext Phil Roberts [mailto:PRoberts@MEGISTO.com]
> Sent: 18. April 2001 12:37
> To: 'mip-qos@research.nokia.com'
> Subject: RE: [MIP-QOS] MIP QoS requirements
> 
> 
> 
> Hi,
> 
> I'd like to interject a few discussion items into what seems 
> to be a most
> productive dialog.
> 
> It's probably worthwhile thinking about which requirements 
> we're discussing
> are MIP specific.  The AAA requirements draft has items which 
> are not MIP
> specific, but not a lot.  This work should stay focused on 
> requirements
> which are closely coupled to MIP and its network model.
> 
> Would you consider keeping the discussion of location privacy 
> separate from
> this for now?  We're not very sure what to do about it in 
> general for MIP.
> 
> Wireless is obviously an area in which this will be used, but MIP has
> managed to avoid being coupled too much to wireless so far 
> and perhaps it
> would be worthwhile decoupling this as much as possible.  
> Perhaps there will
> be a separate effort on QoS characteristics of particular links.
> 
> I haven't seen any mention of the presence of a foreign 
> agent.  Does the
> presence of lack of a foreign agent impact the requirements 
> at all? 

Yes, foreign agent will come into picture a lot, but I think in the solution
space, not the requirements space. What do I mean by that? When there is a
foreign agent, the path change will be limited to the last segment between
FA and MN. We will cover that with the requirement of "localizing the QoS 
(re)programming whenever possible".


> What
> about regional registration?

Given the fact that we should not touch the solution space,  we should leave
this at stating "interoperability with micro-mobility".

> 
> From a requirements point of view can we really say anything 
> more about
> interactions with other protocol sets (such as diffserv, 
> intserv) than allow
> this kind of protocol in the paths between the mobile and the mobility
> agent(s) or correspondents?  And is saying even that too 
> much?  (I know this
> is hard not knowing what the scope of the new working group will be).

I think that saying that will be beyond the scope of our task. The solutions
will say how that is done. Thinking from inside the network, diffserv cloud
need not even be able to differentiate between mobile packet sessions and
fixed packet sessions.

> 
> Can we couch the discussion of fast recovery requirements in 
> terms of the
> failure of an FA/HA?  Otherwise it seems not to be too 
> connected to MIP.

Agreed. We have to add this as a special but relevant case for mobile IP.
The general requirement was "roubustness against topology changes not
related to handover". But FA/HA have special status.

> 
> Phil
> 
> 
>  
>