[rrg] Why won't supporters of Loc/ID Separation (CEE) argue their case?

[Robin Whittle <[e-mail]>]

Hi Tony,

In "Re: [rrg] Recommendation and what happens next" (msg06184) you
didn't respond to the latter part of my msg06181 where I suggested
you and others who support the adoption of a Locator / Identifier
Separation naming structure should write a detailed argument about
why the burden this places on all hosts is justified by savings in
the routing system.  (LISP is misnamed - it does not involve Loc/ID
Separation.)

I am CCing Ran Atkinson, but this isn't just for you and Ran.  It is
for all the people who believe that the RRG should recommend
development and later adoption of a Core-Edge Elimination
architecture: GLI-Split, ILNP, Name-Based Sockets and RANGI.

I am requesting you write detailed arguments for your positions -
including detailed critiques of arguments against CEE (Locator /
Identifier Separation) and of arguments for CES architectures.

CEE proponents are not the only RRG participants who have failed to
adequately argue their case.  The LISP team have not argued why LISP
is superior to other proposals, contributed critiques of other
proposals or responded to critiques of LISP.

No CEE proponent - except perhaps Lixia, who I think may support CEE
in the long-term future - wrote a critique of any of the CES proposals.

The failure of the proponents of one proposal to write detailed,
constructive, critiques of other proposals has been highlighted as a
problem by k claffy (msg06044) and Eliot Lear (msg06054).

I urge you, Ran and other CEE supporters to argue in detail why your
preferred CEE solution is better than any CES solution.  If you like,
please argue against one particular set of arguments for a particular
CES proposal:

   Recommendation suggestion from RW
   http://www.ietf.org/mail-archive/web/rrg/current/msg06162.html

I can't see what constructive purpose is served by you ignoring
suggestions such as this, which I have made before.  If you have
reasons for not wanting to argue the case for architectural changes
you prefer, then I think you should explain them to the RRG.  You you
are expecting us (msg06184) to debate all these issues and reach
consensus on support for a single proposal - yet you have so far
failed to debate your own preferred position.


There has been essentially no response from CEE proponents to my
previous attempts to debate my argument that the burdens CEE places
on all hosts are not justified by the reduced complexity in the
routing system CEE promises, compared to CES.

   Arguments in favour of Core-Edge Elimination vs. Separation?
   http://www.ietf.org/mail-archive/web/rrg/current/msg05745.html  Jan 22

       Only one CEE proponent responded - Xiaohu Xu (msg05759) and
       he did not argue why the extra burdens on all hosts are
       justified.  He admitted to a problem with CEE (Loc/ID
       Separation): the greatly increased difficulty - or perhaps
       impossibility - of filtering incoming packets based on the
       Identifier (currently an IP address) in the source field.


   Today's "IP addr. = ID = Loc" naming model should be retained
   http://www.ietf.org/mail-archive/web/rrg/current/msg05864.html  Feb 2

   CES & CEE are completely different (graphs)
   http://www.ietf.org/mail-archive/web/rrg/current/msg05865.html  Feb 2

       No replies to either, despite me referring to them in
       subsequent messages.


   Vote [1] Ivip, [2] LISP
   http://www.ietf.org/mail-archive/web/rrg/current/msg06144.html  Feb 26

       Contains critique of all CEE proposals: "Locator / Identifier
       Separation naming model would decrease the speed of session
       establishment and unreasonably burden all hosts with extra
       traffic and responsibilities."

       No replies.


   Recommendation suggestion from RW
   http://www.ietf.org/mail-archive/web/rrg/current/msg06162.html

       Contains more detailed arguments against all CEE proposals.

       No replies.



Here is my suggestion again, in summary and in detail:

            As a start to this process, perhaps Tony or others could
            write why CEE and Locator / Identifier Separation in
            general, or ILNP (or another CEE architecture) is
            the best choice over any CES architecture.  I think
            this would include reasons why the extra burdens on
            all hosts and the generally increased delays in
            establishing communications would be justified by not
            having to add so much complexity to the routing system
            as a CES architecture would.

As part of this, perhaps you could write something in favour of CEE
and Loc/ID Separation in general, or in favour of ILNP alone or
whatever - and state why you (and I guess others) believe it is worth
changing host stacks (and apps?) to use this new naming model and so
burdening hosts with more work, more packets, more delays etc. for
the purpose of not having to add so much (or perhaps anything) to the
interdomain routing system (as happens with CES architectures) - and
perhaps for other purposes.  That is for IPv6 only.

Then, if you have any recommendation for IPv4, likewise state what it
is and why it is superior to the alternatives.

At present there is one alternative recommendation for IPv4 and IPv6
- my msg06162.  But if you write something like what I suggest, then
there would be a second alternative recommendation - and hopefully
other people will do the same and contribute their ideal
recommendation text.

You wrote:

>> What I would like to have happen next is nothing but fantasy.

You can keep it a secret fantasy, if you like - which pretty much
ensures it won't happen.

I don't think your position of co-chair should inhibit you from
stating your personal views, hopes, fantasies or whatever.  Indeed I
think being co-chair means you should state these, firstly so
everyone knows what you believe and why - and secondly to set an
example for others to do the same: stating what they believe *and*
why this is a better set of beliefs than the alternatives.

In addition to stating your arguments in a standalone fashion, I
think it is best to quote and respond to detailed arguments against
your preferred positions, and those in favour of contrary positions.

 - Robin