Stefano,
Lots of thanks for a prompt and detailed response.
If, as you say, the SID is just a global index into the local label space of
each LSR, then the statement "PE1 pushes two label on the packet: the top one
is 102" (for SID=102) is grossly misleading IMO, and should be changed.
Regards,
Sasha
Email: [email protected]
Mobile: 054-9266302
> -----Original Message-----
> From: Stefano Previdi (sprevidi) [mailto:[email protected]]
> Sent: Monday, September 29, 2014 3:56 PM
> To: Alexander Vainshtein
> Cc: Loa Andersson; [email protected]; [email protected]; [email protected]; rtg-
> [email protected];
> [email protected];
> <[email protected]>; [email protected]
> Subject: Re: [spring] Routing Directorate QA review of draft-filsfils-spring-
> segment-routing-mpls
>
> Sasha,
>
> SIDs are globally allocated and distributed. So, the answer to your question
> about P1, P2, P3 and P4 knowledge of the SID representing PE2 is "yes".
> Routers within the domain must know the SID representing PE2 (e.g.: 102).
>
> Now, when it comes to the dataplane, it has to be understood that the label
> on top of the packet MAY have local significance and therefore MAY change
> at each hop according to mpls architecture. It is explained (succinctly, I
> admit)
> in section 3.
>
> This because the SID represents an index into a label space, not a label
> value.
> IOW, there is no requirement from SR to have globally assigned labels. The
> SR architecture makes use of globally assigned indexes onto locally scoped
> label space. This is, btw, what been implemented and currently (interop)
> tested.
>
> s.
>
>
> On Sep 29, 2014, at 2:28 PM, Alexander Vainshtein wrote:
> > Loa, and all,
> > I have read the draft and your review on the Wiki page, and I think that
> your question about domain-wide labels is worth detailed discussion.
> > In Section 2 "Illustration", the draft says that:
> > - PE2 advertises (in the IGP) a host address 192.0.2.2/32 with its
> > attached node segment 102
> > - PE1 installs the VPN prefix Z in the appropriate VRF and resolves the
> next-hop onto the node segment 102.
> > Upon receiving a packet from A destined to Z, PE1 pushes two labels onto
> the packet: the top label is 102, the bottom label is LZ.
> > 102 identifies the node segment to PE2 and hence transports the packet
> along the ECMP-aware shortest- path to PE2.
> >
> > To me this strongly suggests that label 102 is understood by P1, P2, P3 and
> P4 as a domain-wide label for an LSP that terminates in PE2. This looks
> different from the downstream label allocation process as defined in RFC
> 3031 and RFC 5331.
> > RFC 5331 clearly states in Section 4 "Upstream Label Assignment" that
> > <quote>
> > If the binding between L and F was made by a third party, say R3,
> > and then advertised to both Ru and Rd, we also refer to the label binding
> as "upstream-assigned".
> > <end quote>
> >
> > In our case it seems that assignment of label 102 and its binding to FEC
> > that
> is represented by Segment 102 is performed by PE2, so P1, P2, P3 and P4
> should all treat this label as upstream-assigned. But this is never mentioned
> in the draft.
> >
> > I concur with you that this requires discussion in the MPLS WG, and I am
> not sure such a discussion should be postponed until the SPRING WG LC.
> >
> > I must admit that I did not read other SPRING documents (including one on
> interoperability with existing label distribution protocols), and hence I can
> be
> wrong in my conclusions. But the MPLS WG looks to me like exactly the place
> where this should be clarified.
> >
> > My 2c,
> > Sasha
> > Email: [email protected]
> > Mobile: 054-9266302
> >
> >> -----Original Message-----
> >> From: spring [mailto:[email protected]] On Behalf Of Loa
> >> Andersson
> >> Sent: Monday, September 29, 2014 1:56 PM
> >> To: [email protected]
> >> Cc: [email protected]; [email protected]; [email protected];
> >> draft-filsfils-spring- [email protected];
> >> <[email protected]>; mpls- [email protected]
> >> Subject: [spring] Routing Directorate QA review of
> >> draft-filsfils-spring- segment-routing-mpls
> >>
> >> Hello,
> >>
> >> I have been selected as the Routing Directorate QA reviewer for
> >> draft-filsfils- spring-segment-routing-mpls-03.txt.
> >>
> >> The Routing Directorate QA reviews are intended to be a support to
> >> improve the quality of RTG Area documents as the pass through the IETF
> process.
> >>
> >> This is the QA review at the time of wg document adoption poll.
> >>
> >> Document: draft-filsfils-spring-segment-routing-mpls-03.txt
> >> Reviewer: Loa Andersson
> >> Review Date: 2014-09-28
> >> Working Group Adoption Poll end date: Oct 8, 2014 Intended Status:
> >> Standards Track
> >>
> >> Please find the review at:
> >>
> >> http://trac.tools.ietf.org/area/rtg/trac/wiki/QA%20review%20Sep%20201
> >> 4
> >>
> >> /Loa
> >>
> >> --
> >>
> >>
> >> Loa Andersson email: [email protected]
> >> Senior MPLS Expert [email protected]
> >> Huawei Technologies (consultant) phone: +46 739 81 21 64
> >>
> >> _______________________________________________
> >> spring mailing list
> >> [email protected]
> >> https://www.ietf.org/mailman/listinfo/spring
> >
> > _______________________________________________
> > spring mailing list
> > [email protected]
> > https://www.ietf.org/mailman/listinfo/spring
_______________________________________________
spring mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/spring
