Re: Consensus on identifier/locator split?

[Alain Durand <Alain.Durand@Sun.COM>]

Are you making the assumption that all your ISPs will route your PI?
If not, you will always end-up with multiple addresses anyway.

To answer Tony's question, I think this split is unavoidable.
The question is where it happens (bellow TCP, in TCP or above)
and how to opt-in. For transition reasons, I think that opt-out is not 
realistic.

   - Alain.



marcelo bagnulo wrote:

> Hi Tony,
>
> I think that a previous question that we may ask is whether we agree that we
> will use PA addresses (topologically meaningfull addresses)  or we will
> assign PI addresses.
>
> If we choose to use PA addresses, then we need to have multiple PA addresses
> in multi-hmed environments, this implies that when a node A is communication
> with a multi-homed node, node A has to know that all the PA addresses belogn
> to the same mh node, so the location and id function are splitted at some
> level (IP, transport, application), because somewhere someone knows that
> multiple locators are linked to a sinlge entity.
>
> If we use PI addresses we only have a single address which accomplishes both
> the id and the loc fucntion so we don't have to split.
>
> So, the question fo me is, do we agree we will use multiple PA addresses in
> multi-homed environments? if so i would say we agree that the identifier
> function and the location function are splitted.
>
> Regards, marcelo
>
> PS: I know smoe are unconfortable with this PA PI terminology, please read
> as topology meaningfull and topology independent
>
>  
> > -----Mensaje original-----
> > De: owner-multi6@ops.ietf.org [mailto:owner-multi6@ops.ietf.org]En
> > nombre de Tony Li
> > Enviado el: miercoles, 06 de agosto de 2003 19:23
> > Para: multi6@ops.ietf.org
> > Asunto: Consensus on identifier/locator split?
> >
> >
> >
> > Folks,
> >
> > I don't mean to disrupt other constructive conversations,
> > so please carry on with those.
> >
> > I'd just like to get a sense from the group about where
> > we are so far.  Do we have consensus about splitting
> > the address into locators and identifiers?  Note that
> > I'm NOT asking about specifics, like "how big", "what
> > mappings exist", "is it secure", etc. Do we agree that
> > we want to go down this path?
> >
> > Silence is not assent...
> >
> > Thanks,
> > Tony
> >
> >    
>
>