Re: [rrg] moving towards recommendation: the current plan

[Scott Brim <scott.brim at gmail.com>]

Flinck, Hannu (NSN - FI/Espoo) allegedly wrote on 11/17/2009 3:27 PM:
> I think that targeting for a single recommended solution would not be
> practical. See for instance what has happened during the years in
> IPv6/IPv4 transition. There is a set of tools for a number of deployment
> scenarios. These tools differ quite a lot on their assumptions. 

This looks right.  I like Lixia's proposal for "evolution" and a
toolkit, although not necessarily using the tools she recommends.

> It looks to me that locator identifier separation is similar to the IPv6
> - IPv4 transition in many ways. It touches address structures, contains
> translation and tunneling aspects etc. 

I'm going to get on the soapbox about this one more time: The point of
locator/identifier separation (LIS) is not necessarily to create new
identifiers.  It's truly separation of Identification from Location --
to stop identification functions from depending on location-dependent
information (like addresses).  Therefore LIS does not _necessarily_
touch address structures, does not need to imply translation or tunnels,
etc.  However, to the extent that you cannot get existing identification
functions (at various layers) free of their addiction to
location-dependent information, routing+addressing may be the likely
place to make up for their shortcomings, and to do that you might do all
of these things.  But LIS does not itself does not require changes to
addresses etc.

Back to the main theme now.

Scott