Hi Guilherme- These examples don't speak to path selection constraints or any particular transport framing constraints. This being said, I did assume that these STPs were compatible (as I wasn't trying to show transport layer function - just path object manipulation) Further, I did not state whether A>B meant that these were adjacent STPs in the topology - in my mind I assumed the A>B was assumed to be a loose hop. I tried to just state some very simple nomenclature for describing path objects, and then make some simple statement about how they can be decomposed and re-combined. A and M (for example) could be very differnet types of STPs, how or why they ended up in the path object wasn't germaine to this issue (IMO). Given A>M, and N>Z, concatenating these two paths requires that M and N be examined. If they are not equivalent topologically then the concatenation A>M>N>Z would imply/require some sort of path construction from M to N was necessary. That path selection would [magically] address your concerns. (:-) If they did turn out to be equivalent points, then the adaptation you ask about would presumably have been done between A and M or between N and Z. Again, I wasn't trying to define how the path was selected, just how to manipulate it after the fact.. Hope this helps (and thanks for reading it close enough to ask the questions - valid questions.) Regard Jerry Guilherme Fernandes wrote:

Wouldn't it be possible for A>M to end be terminated at a different layer (L1) than the entry layer L2 for M>Z, and M not have the adaptation function L1->L2?

In that case, the first implication does not hold, and would need Concatenation to be conditional.

(Sorry if this doesn't follow NSI terminology)

Guilherme

...

Can I propose some connection decomposition and manipulation semantics? Try these:

Definition: A Service Termination/Transit Point "STP" identifies a location in the topology. STPs primarily denote a lowest level component of the topology where a service instance may begin, end, or transit. An STP may also identify a higher level topological construct where a service instance may begin, end, or transit. In the case of a higher level construct, the service instance may touch any sub-component to meet the termination or transit constraint.

Definition: Given STPs A and Z, "A>Z" denotes a unidirectional Path beginning at A and ending at Z;

Rule 1: Equivalence: Given two STPs J and K, If J==K ("J is topologically equivalent to K"), then K==J, and both identify the same location (object) in the topology.

Rule 2: Concatenation: Given STPs A,M,N,,Z, a Connection C1:=(A>M), and a Connection C2:=(N>Z), then C1:C2 == (A>M):(N>Z) == (A>M>N>Z) "C1 concatenated with C2 is equivalent to ..."

Implications:

- If Connection A>M exists, and a Connection M>Z exists, Then by concatenation a Connection A>M>Z exists, and by implication A>Z exists. - Given J==K, then A>J == A>K; and (J>K)==J==K; - Given STPs A, M, N, and Z, If M==N, then (A>M):(N>Z) == A>M>N>Z == A>M>Z ==A>N>Z;

Thoughts? Jerry

_______________________________________________ nsi-wg mailing list nsi-wg@ogf.org http://www.ogf.org/mailman/listinfo/nsi-wg