Re: how mobile do we want to be

[marcelo bagnulo braun <marcelo@it.uc3m.es>]

Hi Dave,


El 18/03/2005, a las 16:59, Dave Crocker escribió:

>  On Fri, 18 Mar 2005 13:24:46 +0100, Brian E Carpenter wrote:
> >   We didn't exclude the discussion. I'm just stating it as my
> >   opinion that the draft charter is correct on this point, and that
> >   this was already clear in the multi6 discussions.
>
> the history of this line of consideration is that it has pretty much 
> always excluded any serious effort to integrate mobility and 
> multihoming. when the initial round of many presentations was done, in 
> multi6, integration with mobility was automatically deferred to 'some 
> other group'.

I am sorry but i disagree with you in this point.

When we begin discussing these issues in multi6, multiple people 
claimed that mobility and multihoming are so similar that they could be 
addressed using similar solutions. Becuase of this reaon, i thourougly 
studied if mip6 could be used to provide mutlihoming support and the 
answer is clearly NO. (see draft-bagnulo-multi6-mnm-00.txt)

so i must disagree with your view of the history of the multi6 here. 
The interaction of mobility techniques and multihoming was discussed 
and presented in multi wg. The problem is that mobilit techniqques are 
simply unable to provide multihoming support.

The reason for this is clear. In multihoming, it is necesarry to 
support the fact that some addresses are unreachable. However, in mip, 
they made the assumption that the home address is always reachable.

> in case no one has noticed, responses to efforts at pursuing such an 
> integration discussion have pretty much always tended to be similar to 
> the responses appearing on this thread.  those responses are automatic 
> rejection, without any interest in seeing whether integration really 
> can make sense. in other words, the tone of this online discussion has 
> had more to do with religion than analysis.

I am sorry but imho this is not the case. As mentioned earlier, a 
pretty complete analysis was done w.r.t. these issues and the 
conclusion was rejecting this approach.

Regards, marcelo

> my original posting, here, was attempting to compare this choice to a 
> decision to have tcp behave very differently in lans than in wans -- 
> in other words, to have two different tcps.  the response, here, has 
> been the equivalent of "lans are different from wans, so it's fine to 
> have a different solution for tcp."
>
> it is certainly true that some definitions of mobility and some 
> definitions of multihoming can permit no useful overlap in the 
> technical solution.
>
> however it is also true that that some definitions permit a very high 
> degree of overlap.  folks seem curiously uninterested in being clear 
> about the reasons for choosing the former, rather than the latter, 
> given the serious long-term costs of having different solutions.
>
>   d/
>   ---
>   Dave Crocker
>   Brandenburg InternetWorking
>   +1.408.246.8253
>   dcrocker  a t ...
>   WE'VE MOVED to:  www.bbiw.net