Re: Progressing the three inter-domain I-Ds

[Dimitri.Papadimitriou@alcatel-lucent.be]

hi adrian - see inline for a couple of clarifications




"Adrian Farrel" <adrian@olddog.co.uk>
Sent by: owner-ccamp@ops.ietf.org
16/01/2007 15:11
Please respond to "Adrian Farrel"
 
        To:     Dimitri PAPADIMITRIOU/BE/ALCATEL@ALCATEL
        cc:     <ccamp@ops.ietf.org>
        Subject:        Re: Progressing the three inter-domain I-Ds


Hi Dimitri,

Thanks for your observations. We will fold them in with our review (unless 

the authors want to spin a new version first?)

You wrote:

> o) a couple of specific comments
>
>end of section 2:
>

Missing comment?


[dp] yes the following sentence

"the appropriate technique must be driven by the set of
   requirements for the paths attributes and the applicability to a
   particular technique with respect to the deployment scenario"

the beginning of the sentence is not recommending an approach, 
the "must" as indicated becomes authoritative, this sentence is
cyclic and reads 

"the appropriate technique must be driven by the applicability 
to a particular technique with respect to the deployment scenario"

is the term "path attribute" referring to any kind of LSP attr. ?
if yes please state so, if not please use the correct term ?



With regard to the discussion about advertising inter-AS TE links, you are 

right that this needs clarification. I also feel that this I-D is not the 
place to do it. *If* IGP advertisement of inter-AS TE links is supported, 
it 
will need documentation and this I-D can point to it.

[dp] i would agree that the reasoning throughout the draft is deductive
and the proposed method defined for MPLS-TE (how does this works for bi-
directional LSP for inst. when the inter-domain can be crossed by both
bi and unidirectional LSPs) is put in a path.comp context 

[dp] i mean here that beyond the specific per-domain operations the only
protocol addition is an OSPF one (i leave this question to whether needs
for a separate i-d is more appropriate, but the level of detail must be
the same in both cases, not because embedded in this i-d that accurate
description of processing shall be skipped)

Adrian