[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: No Upstream Uni-directional LSPs?
ERO label subobjects will be one with the U-bit set and one NULL label
without the U-bit set. I think ERO label subobjects are not mandatory though.
By the way, if the last hop subobject in an ERO is followed by
label subobjects, are the label subobjects used for "LSP splicing"?
Also I should try to improve the description around hierarchical LSP
considerations including the deletion process.
At 02/09/06 03:32, Michael I Mandelberg(Isaac) wrote:
>I have given a brief look at this draft. I think the ideas are interesting,
>but more work needs to be done to make it compatible with current rsvp-te
>work. One thing that is missing is a modification to the ERO object. The
>current draft mandates that there is either two label subobjects, one with
>the U-bit set, or else only one, without the U-bit set.
>Sent: Thursday, September 05, 2002 8:22 AM
>To: MATSUURA Nobuaki
>Cc: Michael I Mandelberg(Isaac); firstname.lastname@example.org
>Subject: Re: No Upstream Uni-directional LSPs?
>from the i-d perspective most of the content deals with
>setup, it would also be good if some words concerning
>the deletion can be proposed (i mean the reference to
>the method described in current gmpls signalling i-d)
>and also to clarify the following paragraph wrt to the
>"If a new FA-LSP is required to be set up between the
>LSR and the other edge of the region, the LSR initiates
>the setup of a new reserve-directional FA-LSP. At the
>same time, the LSR may send the Path/Request message
>for the original reverse-directional LSP to the other
>edge of the region."
>another point is as follows: we have implementations
>ongoing within our community for the set of features that
>are described in the common control gmpls signalling i-d's
>-> therefore it would be of great interest to gather more
>information on deployment/utilisation (ie experimental
>i-ds) in the near future in order to be capable to assess
>what might be further considered from that persceptive
>for these i-d's - i think this may also be applied to
>other items that may come out in the same register -
>MATSUURA Nobuaki wrote:
>> Hi Michael,
>> Current GMPLS specifications don't support the upstream unidirectional
>> But MPLS in principle doesn't restrict the direction of an LSP setup.
>> Therefore I think it is a reasonable generalization and posted an I-D for
>> Please refer,
>> (I'm going to revise this draft considering the discussion on the list.)
>> At 02/09/05 12:57, Michael I Mandelberg(Isaac) wrote:
>> >It is my impression that GMPLS does not support unidirectional LSPs with
>> >upstream data flow (at least not RSVP). Is that correct?
>> >Michael Mandelberg
>> >FirstWave SION
>E-mail : email@example.com
>E-mail : firstname.lastname@example.org
>Public : http://psg.com/~dpapadimitriou/
>Address: Alcatel - Optical NA (CTO), Fr. Wellesplein, 1
> B-2018 Antwerpen, Belgium
>Phone: Work: +32 3 2408491 - Home: +32 2 3434361