Re: [rrg] A Revised critique for LISP - 745 words

[Robin Whittle <[e-mail]>]

Hi Lixia,

My position regarding critiques is that it is unrealistic of you and
Tony to combine or rewrite parts of the texts of multiple critiques and
expect the authors of those critiques to be happy with the result.  If
you didn't have a 500 word limit, then two or more critiques could be
included verbatim - which would be the best approach, I think.

Your rewritten critique:

  http://www.ietf.org/mail-archive/web/rrg/current/msg06032.html

has 745 words.  If you would allow another 255 words, then you could
have both my critique and Noel's, as cohesive, independent pieces of
text - and keep me and I think Noel happy.

Your new critique does not cover all the things I mention in mine:

  http://www.ietf.org/mail-archive/web/rrg/current/msg05728.html

If you only want a single critique, then I would still prefer mine to be
in the RRG Report than either Noel's or your rewrite of Noel's - but
what I really want is Noel's and mine to be in the report, verbatim, on
equal terms.

I don't think it would be a good use of time to debate most of the text
you have created - though below I concur with Dino's concerns about the
last paragraph.

In an ID:

  http://tools.ietf.org/html/draft-whittle-rrg-critiques

(currently just a place-holder) I will include full length versions of
all the critiques, including Noel's which do not appear in full in the
RRG Report.

   Abstract

   This ID will contain the full-length versions of all critiques
   which were not included in full in the RRG Report.  It may also
   point to discussions I consider significant, and include any
   "rebuttals" or "reflections" which were not included in full in
   the Report.  This ID is to support and document the IRTF Routing
   Research Group discussions concerning scalable routing, in early
   2010.  It is not a formal product of the RRG.  To what extent
   its contents are supported by RRG consensus can only be
   determined by the co-chairs.


Dino wrote:

>> Last but not least: One would not be able to see global routing table
>> size reduction unless/until LISP has been adopted by significant
>> number of networks. On the other hand, LISP is potentially a useful
>> tool in data centers where its one-level of indirection may help
>> significantly simplify the support for virtual servers.
> 
> This is simply not true. The more specifics from site advertisements are
> withdrawn at the expense of coarse prefixes being injected by PITRs. The
> difference in the number of 1 or 2 orders of magnitude.

Yes, but I think factors of ~10 or ~100 are an underestimate.

These "coarse prefixes being injected by PITRs" have no specific name in
LISP, but I call the same things "Mapped Address Blocks" (MABs) in Ivip.
 I expect IPv4 MABs to be /24s to /16s or perhaps /12s.  Since many
end-user networks only need one or a handful of IPv4 addresses, I think
each MAB could serve the needs of thousands of end-user networks.  I
think the same applies to LISP, so I would say for LISP and Ivip that
there will generally be factors of 10^3 to 10^4 ratios between the
number of end-user prefixes supplied and the number of covering
("coarse" / MAB) prefixes advertised in the DFZ.

It is a feature of a CES architecture such as LISP or Ivip - both of
which support portability, multihoming and inbound TE benefits for all
packets irrespective of the level of adoption - that the scaling
benefits are in direct proportion to the level of adoption.  So I think
your first sentence is incorrect and any such statement does not apply
to LISP or Ivip.

Please see:

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

The routing scaling benefits resulting from LISP and Ivip will be in
direct proportion to adoption.  There is no threshold in the level of
adoption - as implied by your use of "significant number of networks" -
at which routing scaling benefits begin

            ^
         B  |        *      Core-Edge Separation (CES)
         e  |       **
         n  |      ***
         e  |     ****
         f  |    *****
         i  |   ******
         t  |  *******
            | ********
            |*********
            0--------->
              Effort ~= level of adoption


Dino also wrote:

> And why is there two points in the same paragraph? The data center
> comment came out of no where.

Indeed.  What is a "data center"?  I have no idea what this second
sentence means.  It doesn't seem to be a critique of LISP.

  - Robin