Hi Filippo,
The reason for the RRO inclusion in PCReq in draft-ietf-pce-pcep-12.txt is because the PCE is stateless. As far as I know, all the PCE drafts and RFCs assume that PCE is stateless. Thanks for your review on the latest GCO draft, I will make some corrections to make it consistent. Regards, Young _____ From: Filippo Cugini [mailto:[EMAIL PROTECTED] Sent: Thursday, June 26, 2008 5:24 AM To: Young Lee Cc: [email protected] Subject: Re: [Pce] PCE GCO update Hi, please see in-line. 1) in case of stateful PCE, is it mandatory to include the full list of RROs in the PCReq message? Young>> I am not sure of your definition of stateful PCE. If you mean stateful PCE to be the PCE that keeps the state of the active LSPs at the PCE, this is beyond the scope of PCEP in IETF. Nonetheless, if this is what you mean by stateful PCE, I am not sure why you need to send the full list of RROs in the PCReq message since the PCE keeps such information with stateful PCE. When you want to do re-optimization with stateless PCE, you need to send the full list of RROs of the LSPs to be re-optimized in the PC Req message. This way PCE needs not maintain the state of LSPs at the PCE level. Filippo>> agree that in case of Stateful PCE (yes, the one that keeps the state of active LSPs) there is no need to send the full list, however, in draft-ietf-pce-pcep-12.txt, section 7.4.1 and .2, it is reported that in case of re-optimization, PCReq messages MUST include RROs. 2) In section 2, Terminology, GCO is defined to consider only the entire network, while in the rest of the document also network subsets are allowed. Is the use of "Global" associated to both the cases of entire network and network subsets or is it just related to the entire network? Young>> Although GCO considers the entire network in general case, there are applications of GCO where only a subset of TE LSPs can be re-optimized. This has a protocol support. I believe you are quoting the statement in the Introduction, "Alternatively, the application may consider a subset of the LSPs and/or a subset of the network topology." I may have to delete the last clause. This can be achieved, but we don't have ways to limit the whole topology into a subset in the current PCEP for GCO. Are you aware of any good application for this case? Filippo>> Yes, exactly that statement in the introduction which should become consistent with the rest of the document. To restrict the computation to a network subset, a possible solution could be the use of IRO object, e.g. specifying theIP prefix or AS number (don't know how to indicate Area ID). Typo: in section 4, "section 6.1.14 of [RFC4657]" -> "section 5.1.14 of.." Thanks Filippo
_______________________________________________ Pce mailing list [email protected] https://www.ietf.org/mailman/listinfo/pce
