[rrg] Deadline for critiques - and RANGER

[Robin Whittle <[e-mail]>]

Short version:   I think we need another two weeks to write a proper
                 set of critiques.  I think writing the "rebuttals"
                 and the "reflections" won't take long - maybe a
                 week to do both.

                 We are deciding how to advise the IETF on the best
                 way to develop a once in several decades upgrade on
                 the most widely used, entrenched, IT system in the
                 world.

                 No-one knows how much is invested in routers and
                 the, Internet aspects of host OSes applications, but
                 it is clearly in the tens or hundreds of billions of
                 dollars.

                 The upgrade has to last for decades, be performed
                 while it is running and be widely adopted on a
                 purely voluntary basis.  (Also, I think, the upgrade
                 needs to support billions of mobile hosts - which is
                 surely the future of the devices currently known as
                 "cellphones".  I think this could best be done via
                 global mobility of their address space, no matter
                 what access network they are connected to.)

                 With two days to go, this "critique" phase of the
                 RRG's work has only produced results for 2 of 16
                 proposals.  We need more time.



Hi Lixia and Tony,

I support the late inclusion of Fred Templin's RANGER core-edge
separation proposal.  I would like to contribute to a critique, but
there's no way I can do this in the next week or so.  It is a
well-documented architecture so there is a lot of material to read.


I would like you to consider a significant extension of the deadline
for the critique phase of this process - to Monday 1st February.  I
would be suggesting this anyway, but now we have one more,
extensively documented and well-researched proposal to read and consider.

Further to msg05670, here is my understanding of the Critique process
for each of the now 16 proposals:

   One or more people write a critique and post it to the list.

   Other people may try to contribute to that critique (via the
   list or privately) and the original author(s) may or may not
   accept such alterations to their satisfaction.  So updated
   versions of the original draft of the critique may be posted to
   the list.

   Multiple critiques from multiple people or groups of people may be
   posted to the list in this manner, being revised by their authors
   - and I guess copied in part to be the basis of other critiques.

   At deadline, if there are two or more competing critiques which
   still have support, then Lixia and Tony will choose one.

The critiques are a tricky thing.  Some of people (me at least) want
to contribute to multiple critiques, so we have a lot of work to do
by the deadline.

Since various people may want to contribute to the one critique, we
need to perform autodiscovery and then work together.  Autodiscovery
really needs to be done on-list with an announcement of intention to
contribute to a critique.  Otherwise how could anyone know who to
work with?

My understanding is that the deadline is Monday 18th January (Lixia's
message msg05648).

Due to misunderstandings of the original instructions (msg05544, with
a 5th January deadline) - which I misunderstood - two proposals have
had an "analysis" (now "critique" msg05664) written by the
proponents:  Name Based Sockets and Name Overlay Service.  But
critiques need to be critical and written by someone other than the
proponents.

Here's my understanding of the state of the critiques:

  ILNP
    http://www.ietf.org/mail-archive/web/rrg/current/msg05539.html
    Ran Atkinson

    Joel Halpern and Yakov Rekhter wrote a critique (msg05624).
    I support their critique but in the future I may have other
    things to mention about the need for DNS lookups to achieve
    equivalent certainty to today's protocols that the initial
    response packet really goes to the host which the request
    supposedly was sent by.  I need more time to work on this
    and hope to do it by the time of the final choice discussion.


  LISP
    http://www.ietf.org/mail-archive/web/rrg/current/msg05503.html
    Vince Fuller, Dino Farrinacci, David Meyer and Darrel Lewis.

    I announced I would contribute to a critique of LISP.  So far
    no-one has indicated that they would also contribute.


  TIDR
    http://www.ietf.org/mail-archive/web/rrg/current/msg05538.html
    Juan Jose Adan

    I announced I would contribute to a critique of TIDR.  So far
    no-one has indicated that they would also contribute.

    However, Mohamed Boucadair has sent Juanjo and I a copy of the
    TIDR ID with his extensive thoughtful comments.  I hope that he
    or Juanjo will post it to the RRG list as an attachment to
    contribute to the RRG discussion of TIDR.

  Name Based Sockets
    http://www.ietf.org/mail-archive/web/rrg/current/msg05543.html
    Christian Vogt

    I announced I would contribute to a critique of Name Based
    Sockets.  So far no-one has indicated that they would also
    contribute.

  Ivip
    http://www.ietf.org/mail-archive/web/rrg/current/msg05533.html
    Robin Whittle

    Christian Vogt wrote to me offering to coordinate a critique of
    Ivip - which I very much appreciate.  I suggested he announce
    this on the list.  I think the choice of coordinator should be
    made by whoever wants to contribute.

    I think it is vital that the LISP folks state their objections to
    the other core-edge separation proposals, including Ivip, TIDR
    and now RANGER.

  RANGER
    http://www.ietf.org/mail-archive/web/rrg/current/msg05665.html
    Fred Templin

    I have offered to contribute to a critique, but not by the the
    current deadline.

To date, only 2 of 15 proposals have critiques.

For a further 4 proposals, someone (only me so far) has announced
their intention to contribute to a critique.

As far as I know, there has been no announcement that anyone is
interested in writing critiques for the following 9 contributions.
Those marked * are the ones, which in my opinion (msg05562) do not
constitute actual proposals for the scalable routing problem.
Nonetheless, they are part of the process and need critiques (msg05564).

*  2-phased mapping for Internet core/edge split schemes
     http://www.ietf.org/mail-archive/web/rrg/current/msg05536.html
     Wei Zhang

*  Aggregation with Increasing Scopes: An Evolutionary Path Towards
   Global Routing Scalability
     http://www.ietf.org/mail-archive/web/rrg/current/msg05542.html
     Dan Jen, Dan Massey, Robert Raszuk, Lan Wang, Xiaohu Xu,
     Beichuan Zhang and Lixia Zhang.

*  Enhanced Efficiency of Mapping Distribution Protocols in
   Map-and-Encap Schemes
     http://www.ietf.org/mail-archive/web/rrg/current/msg05540.html
     K. Sriram, Young-Tak Kim, and Doug Montgomery

   Global Locator, Local Locator, and Identifier Split (GLI-Split)
     http://www.ietf.org/mail-archive/web/rrg/current/msg05537.html
     Michael Menth, Matthias Hartmann and Dominik Klein

   hIPv4
     http://www.ietf.org/mail-archive/web/rrg/current/msg05529.html
     Patrick Frejborg

*  Layered Mapping System
     http://www.ietf.org/mail-archive/web/rrg/current/msg05534.html
     Sun Letong, YinXia, Wang ZhiLiang, Wu Jianping

*  Mapping system based on compact routing
     http://www.ietf.org/mail-archive/web/rrg/current/msg05519.html
     http://www.ietf.org/mail-archive/web/rrg/current/msg05531.html
     Hannu Flinck

   Name overlay (NOL) service for scalable Internet routing
     http://www.ietf.org/mail-archive/web/rrg/current/msg05532.html
     Yangyang Wang

   RANGI
    http://www.ietf.org/mail-archive/web/rrg/current/msg05505.html
    Xiaohu XU


Here are some arguments for a deadline extension, such as to Monday 1
February.

In your original deadline (msg05544), now extended twice to Monday 18
January seems to have been based on an underestimate of how long it
would take people to read proposals, nominate themselves on the list
and actually write and refine the critiques.  Only Joel Halpern and
Yakov Rekhter were able to write a critique by the first revision of
the deadline (msg05556).

There are four scalable routing proposals so far with no-one
nominating to write a critique: GLI, hIPv4, NOL and RANGI.  I plan to
write critiques of these in time to contribute to the RRG's final
choice discussion - but not in the next week or so.

My first priority is to make some improvement to Ivip's fast push
mapping system, to respond on list to Mohamed Boucadair's comments on
Ivip.  Then I am going to contribute to critiques of TIDR, LISP, Name
Based Sockets and RANGER.

We need more time for people to write a proper set of critiques.

I recall your initial timetable was for 14 days between stages - 14
days for the proponents to write a "rebuttal" of the critique, and
then 14 days for people other than the proponents (presumably those
who wrote the critique) to write a "reflection".

I think this is a good process, but that you need to give us more
time for the critique phase.

I can't speak for other proponents, but it won't take me 14 days to
write a 500 word "rebuttal" of the critique Christian Vogt and
hopefully others will write for Ivip.  (Not that my response will
necessarily be a simple rebuttal.)  The "rebuttal" stage is an
entirely parallel process, since each proponent only has to write one.

Some folks want to contribute to multiple critiques, which is a
serial process.  Also, critiques involve reading a potentially large
body of material which we are not familiar with and then carefully
crafting a respectful and well-informed critique in only 500 words.
Ideally, this will also involve working with other contributors and
checking with the proponents that we haven't misunderstood something.

The proponents, when responding to the chosen critique, already know
their proposal inside out and are responding to only 500 words, which
they have seen in various drafts on the list prior to the deadline.

I don't think it will take the authors of the critique - and anyone
else who has something to say - 14 days to write a "reflection".
They have already read the proposals and are only responding to 500
words.

I think it will only take a week or so to do both the "rebuttal" and
"reflection" stages.

 - Robin