[Idr] Comments on draft-ietf-idr-bgpls-inter-as-topology-ext-01

["Ketan Talaulikar (ketant)" <[e-mail]>]

Hello Authors,



Thanks for this draft update and presenting it at the IDR session on Monday.



I refer to our exchange<https://mailarchive.ietf.org/arch/msg/idr/42e5k0TwR6RKrJ9QiiKVfmwlLSE> during the WG adoption call on some issues/concerns on this document and also some comments on the mailing list on the previous versions of this draft from Acee, Jeff and others.



I see that the draft still continues to carry this notion of "native IP" scenario and the wrong premise of redistribution of the prefixes corresponding to the inter-AS links being used to "describe" such links via BGP-LS. I have no objection if any provider wants to use the BGP-LS information in this way - after all what a consumer does with the BGP-LS information it is outside the scope of BGP-LS. However, I continue to have objections to this entire section 3.1 continuing in this WG draft. Distribution of external prefixes (via redistribution) is already covered in RFC7752. There is nothing new here. What is new is this assumption of interpreting redistributed prefixes as inter-AS links which is wrong and MUST never be generalized as such in a standards track RFC.



Can you please remove this entire Section 3.1?



The rest of this document covers the important gap that was left by https://tools.ietf.org/html/rfc7752#section-3.5 and hence my support for this document.



I see that you have defined a new Link-State NLRI type for describing an Inter-AS link. I understand your motivation for doing this since we don't have Remote Node Descriptors for such links. I would suggest that instead of making it specific to Inter-AS scenario, we can view it as a "stub" link in a generic manner. This way, it may be also used for other scenarios where we don't have a IGP adjacency or it's part of the IGP adjacency but running in passive mode.



We can discuss further on this NLRI type once we have an agreement on above.



Thanks,

Ketan