The MPLS WG Archive

[neil.2.harrison@bt.com]

Hi Alex,

	You asked:
> > -       "server layer trails (in an OTN) = client layer links", and
> networks
> > operators *will* have to support multiple client layers for a very long
> > time...including some large BW servives directly off the L1 fabric.
> 
> Not sure I follow the logic here. Can you elaborate a bit.
> 
	This is a really important point.  It is a fundamental
characteristic of a layered network architecture.  It (and other functional
modelling concepts of layered networks) are more fully described in G.805
(ITU) and also in some text books like:
	-	'Global Information Networking' Varma et al. Artech
	-	Broadband Networking' Reid & Sexton. Artech

	In essence, if I want to create a topology at some arbitrary layer
(in some technology) I can do this by connecting boxes with links to form
some sort of graph......this is the simplistic box diagram approach.  But
each of these links can be provided by a lower layer network, which is
itself a topology of boxes and links.  And this process recurses until we
hit the 'duct'....which is the lowest layer network of all (and which, BTW,
determines a bound on the inherited availability performance of all higher
layers).

	Hence, a link connection between 2 nodes at layer N is a small
segment of a large trail at layer N.  The trail end points in layer N are
closely bound to the access points which are the addressable entities of
layer N, ie the entities between which we need to determine routing between
at layer N.  But *each* of the link connections at layer N will generally be
provided by a trail at Layer N-1.  Hence, the addressable access points of
the trail at layer N-1 are clearly not the same as those at Layer N.  I have
tried to show this nested inter-layer trail relationship below (0 = trail
termination point/addressable location of layer network, (X) = switching
point in layer network):

	<----------------------------------Trail at layer
N------------------------>
	
0-----------(X)---------------(X)-----------------------(X)----------------0
	                                        /   <--------------------->
\
	                                     /  link connection at N      \
	                                   /
\
	
0--------(X)---------(X)-----------0
	                                  	<----trail at
N-1--------------->
							           /
\
	                                            /                      \
	                                          /
\
	                                         0  -----(X)----(X)----0
	                                         <---trails at N-2----->
	                                                    /           \
	                                                  /                \
	                                                    etc to duct

	BTW nested LSPs also fit this model.......but this has not been
recognised yet.

	Neil