Hi all,
I have couple questions related to Link ID of inter-AS link. According to the document it is the Router ID of Remote ASBR.
To be precise, the I-D says "TE Router ID". I think the difference is very important.
[ZRH]Yes, and this can be easily got by manual configuration in the local ASBR.My question: - as I understand the mechanism of how TE information for links R6-R9 and R8-R10 is populated into R5's TE DB is outside the scope of this document. I this mechanism is not IGP but EGP (BGP), then what is the benefit to introduce new LSA in IGP (OSPF) and not use exisitng mechanisms.
I think you mean R8-R9 (at least in the latest revision of the I-D).When you say an "existing mechanism" I don't think there is any mechanism for advertising TE information in BGP although there are proposed extensions. But there would be two issues:
1. How to stop the information propagating further than AS2 (AS1 does not want/need to know about the inter-AS links between AS2 and AS3). I guess we could use the new AS_PATHLIMIT stuff.
2. How to get the information into the TED of every LSR in AS2 without them all having to be BGP speakers. I think this is why the IGP is more appealing. The question arrises (as indicated by Renhai, below) as to whether this information only needs to reach PCEs, PCEs and ASBR, or all LSRs.
[ZRH]I think BGP could also be another alternative especially for PCE scenario, but it's not very applicable to per-domian method because the inter-AS TE information may be needed throughout the local AS in case that the path computation function is distributed among LSRs in the local AS, and BGP can hardly advertise the TE information among those LSRs if BGP sessions are not established between the ASBR and those LSRs. So IGP extension is considered in this draft.
[SNIP]
PS. I think that OSPF WG needs to be informed as well.
Yes. If CCAMP and/or PCE think that this is something they want to pursue, then we will definitely need to discuss this with the OSPF working group. But I don't want to waste their time or energy before we have discussed this a bit more.
Thanks, Adrian