The MPLS WG Archive

["Kullberg, Alan" <akullber@netplane.com>]

Hi Kireeti,

Thanks for the response.  The "variant of LMP" to which you
refer sounds overly complicated for the problem to be solved.
Perhaps that's why you are seeking an alternate solution.
Does the solution involve a new Object/TLV to travel in the
PATH/REQUEST message to aid in communicating to LSR-B the binding
between the FA-LSP and the PATH/REQUEST?  In my earlier email,
I suggested that the 6 octet LSPID (from CR-LDP) could serve
this purpose just fine.  To be specific, in the PATH/REQUEST
message from LSR-A to LSR-B that is requesting timeslot(s) be
assigned from the FA-LSP (lambda), a new TLV is included that
carries the LSPID corresponding to the FA-LSP that is to be
used for stacking.

Alan

> -----Original Message-----
> From: Kireeti Kompella [mailto:kireeti@juniper.net]
> Sent: Wednesday, August 16, 2000 5:17 PM
> To: akullber@netplane.com; kireeti@juniper.net; mpls@UU.NET
> Subject: RE: generalized MPLS signaling question
> 
> 
> Hi Alan,
> 
> > Let's assume the lambda-switched paths are unnumbered and 
> are assigned
> > "interface" indices as per the unnumbered draft.  When 
> LSR-B receives
> > an ERO with the "Interface ID" subobject, how is the 
> association made
> > with the correct lambda-switched path for timeslot allocation?
> 
> Excellent question.  The implicit assumption was that with out-of-band
> signalling, some other means of identifying the links is used (e.g., a
> variant of LMP).  I.e., LSR-A knows (per the ERO) which FA
> (lambda-switched path) to use.  LSR-A and LSR-B have already 
> identified
> the links (e.g., with some variant of LMP).  LSR-A then sends the Path
> message with a Link ID object (from the bundling document) to 
> let LSR-B
> know which FA LSR-A wants to use.
> 
> However, this is not altogether satisfactory (variant of LMP, etc).
> An alternate solution is being worked on.
> 
> Kireeti.
>