Short version: Noel seems to have asserted that my critique of LISP
is inadequate - due to being focused on minor
details rather than real architectural problems.
Is he suggesting there should be a critique of LISP
being a Core-Edge Separation architecture? If so,
please see my next message.
Hi Noel,
>From our recent exchanges I thought you viewed my LISP critique as
being in some way unreasonable - and that this was the basis for your
concern that it might be viewed as some kind of official RRG
assessment of LISP.
Now, in msg05739, you wrote that in your view my critique has:
> major problems (too much focus on minor, passing problems, and not
> enough attention to unavoidable architectural limitations), I will
> prepare an alternative, and you and Lixia can pick whichever seems
> best.
so I think you consider my critique as being too *weak*.
You don't support your assertion with any details.
Unless you can really change my mind about LISP, I am unwilling to
try to accommodate your concerns in the critique I wrote. I have
only one word left and I don't want to take anything out.
I hope you will write a critique of LISP which you are happy with.
Likewise anyone else who thinks I missed something or was
unreasonable in some way.
Eliot suggested that all critiques be included in the RRG report. As
long as they are reasonably well written, and don't overlap too much,
I support this.
So far, no more than one critique has been written for each proposal,
though I haven't yet worked on RANGER, and Lixia and I think the ILNP
critique should go further.
Tony, being mathematically inclined, reacted to the prospect of there
being two or perhaps even more critiques for a single proposal as
being a step too far towards the dangers of Infinity.
If you want your critique of LISP to go in the RRG Report, then I
will support this inclusion.
Here's another mathematical perspective. The LISP team were allowed
1000 words to help RRG folks understand their proposal - but they
used only 206. So perhaps Lixia and Tony could accept further
critiques for LISP to a maximum of 794 words without breaking LISP's
share of the space budget.
With a less restrictive word limit, I would have mentioned more about
LISP - more positive things. I would have noted that it is a
Core-Edge Separation (CEE) scheme and that this is a very good thing
for various reasons, compared to the alternatives: Core-Edge
Elimination (CEE) or attempting to soup up the existing interdomain
routing system with improved hardware, some new BGP version, or
something which replaces the current BGP-based DFZ.
In my view, the LISP-folks (and the designers of APT, TRRP, TIDR and
probably RANGER) took the correct fork in the architectural road by
choosing a Core-Edge Separation architecture. My arguments for why
are in section 4.1 of:
http://tools.ietf.org/html/draft-whittle-ivip-arch-03
If your critique of LISP is in part that it is a Core-Edge Separation
architecture, then I think it would be best to write up a general
critique of CES architectures, and therefore, I assume a general
argument for the superiority of at least one CEE architecture. That
is the subject of my next message.
- Robin
_______________________________________________
rrg mailing list
rrg@irtf.org
http://www.irtf.org/mailman/listinfo/rrg
