The MPLS WG Archive

[Bijan Jabbari <bjabbari@osf1.gmu.edu>]

  This topic was also dealt with to a good extent at the MPLS 2000
conference. We are in the process of making all the presentations
and tutorials at this conference available to everyone. This will
happen most likely as soon as the semester is over here at GMU, ie,
next week. The web address is: 

  http://www.ail.gmu.edu/MPLS2000/default.htm
 
  - Bijan Jabbari


On Tue, 12 Dec 2000, Peter Ashwood-Smith wrote:

> 
> You may find some of the following presentations useful. There is also an
> MPLS tutorial floating around that has some nice stuff on GMPLS. Hopefully
> one of the authors will read this and give you a pointer to it.
> 	
> 	
> http://www.nortelnetworks.com/products/announcements/mpls/events/2000conf.ht
> ml
> <http://www.nortelnetworks.com/products/announcements/mpls/events/2000conf.h
> tml> 
> 
> 
> *	
> *	Greetings!!
> 		> 
> 		> I am a newbie to GMPLS. Was wondering if somebody
> 		> can point me to some documents (not internet-drafts)
> 		> which has some pictures in it. Currently I have
> 		> gone through the generallized signalling document
> 		> which talks about extension of classical MPLS architecture
> even for TDM frame switching,
> 		> wavelength switchinging, etc, but could not get clear
> 		> understanding of the even the basic thing. Can somebody
> 		> comment on the following?
> 		> 
> 		> Once the GMPLS is done. And there is an end to end
> 		> 'LSP' setuped using optical wavelength. Is it that once
> 		> a packet is sent on at the first node using the first
> 		> label i.e. using the first wavelength, it doesn't go
> *	up to the MPLS layer at the intermediate nodes.
> 
> 
> 	It is important to keep in mind that GMPLS creates circuits that can
> be naturally stacked or spliced depending on the goal of the network
> operators. One can create a complete edge to edge TDM circuit of say OC-48,
> 12 etc. and then run IP/POS over that. That TDM circuit could traverse a
> number of  TDM-LSRs that inter-connect with OC-192 trunks and allocate 48
> timeslots (label) to the connection as it traverses the LSR. The circuit
> could also traverse pure optical LSRS that would allocate an entire
> wavelength to the OC-48, and switch it through using mirrors/bubbles or
> whatever. 
> 	The entire IP/POS circuit could then be running MPLS with shim
> headers, or the POS circuit could be used to carry non IP traffic. All of
> this equipment could be running as part of the same network (peer model), or
> the TDM/Optical stuff could be running in its own, completely isolated
> network (overlay model).
> 
> 	As with vanilla MPLS, reaching the end of one technology, does not
> necessarily mean the packet has to go up a level for processing if there is
> a way to map one technology directly to the next.
> 
> 	The little example I just gave is probably the first application of
> GMPLS, as the control mechanism for the core of a ASON style TDM/Optically
> switched network with IP/POS at the edge. There are many other 'possible'
> uses of GMPLS some of which are pie in the sky(today) and others of which
> are more pragmatic.
> 
> 
> 	Hope this helps,
> 
> 	Peter Ashwood-Smith
>