Re: [CCAMP] draft-ali-ccamp-lsp-inquiry-00

["Adrian Farrel" <[e-mail]>]

Right, Khuzema, you caught the right piece of text.
 
The point here is that the LSP is *not* set up in the data plane if it has a
different label. All that is happening is that the LSP-ID and state is being set
up in the control plane.
 
I may have said it before <snarkle> but it is massively helpful for people to
pay close attention to whether they mean "control plane LSP" or "data plane LSP"
when they say "LSP".
 
A
 
From: [e-mail] [mailto:[e-mail]] On Behalf Of
Khuzema Pithewan
Sent: 30 July 2013 15:25
To: CCAMP ([e-mail])
Subject: [CCAMP] draft-ali-ccamp-lsp-inquiry-00
 
Authors,
 
The draft relies on ability to setup GMPLS lsp without committing resources in
dataplane.
 
Only reference I found to setup pre-planned GMPLS is in RFC 6001.
 
The green highlighted part says it is not possible to support 0 bandwidth lsp
for TDM/LSC network. While red part alludes that it can be done.
 
Also I couldn' locate the text in any RFC that describes the NULL label behavior
in GMPLS context.
 
 
RFC6001 5.2.2 says
 
......snip.......
 
However, mechanisms for provisioning (pre-planned or not) a TDM or
   LSC LSP with 0 bandwidth is currently not possible because the
   exchanged label value is tightly coupled with resource allocation
   during LSP signaling (e.g., see [RFC4606] for a SONET/SDH LSP).  For
   TDM and LSC LSP, a NULL Label value is used to prevent resource
   allocation at the data plane level.  In these cases, upon LSP
   resource commitment, actual label value exchange is performed to
   commit allocation of timeslots/ wavelengths.
 
.........snip........
 
 
Khuzema