[Nsi-wg] NSI - Call Issues

[John Vollbrecht]

We will have a call tomorrow at 9ET.  I suggest that this be the last  
call this year and that we restart calls either Jan 6 or Jan 13.  We  
can continue exchanges on skype and by email during this time.

I would like to try to have a skeleton document to start discussing by  
the middle of January and a document for OGF ready by early February.   
To do this we will have to make some assignments tomorrow or soon  
after, and probably work in smaller groups to come up with wording.

I have two issues that need to be decided for the document.

1) Naming of two sides of NSI interface.  We originally named these  
the requestor agent and network service agent.  This after significant  
discussion, especially about the fact that we should not use provider  
because it had connotations of the commercial providers.  Later we  
decided to go ahead and use the name provider because Network Service  
Agent was confusing with other names like Network Service Actor.  Also  
requestor and provider are terms used by others so seemed reasonable.   
The person who was most strongly against using the name provider was  
not on the call when we made the decision to change, and he continues  
to feel strongly that it is a bad name.

I think we made a decision not to use provider and should not change  
it without agreement from the parties involved in the decision.   
Therefore I would like to change the name of different sides of the  
NSI to something else.  Suggestions are a) requester agent- service  
agents,  b) client  agent - server agents.

We need to determine the name in order to make the document.  I think  
this is a NSI group decision, and I am hoping we can decide tomorrow.

2) There is an issue with naming end points on the transport resource  
controlled by an NSA.  To explain-

A NS Service agent  deals with connections between ports.   In NML  
terms the ports are part of a group of connected links and nodes.  The  
edge of the group might be a node or a link.  The problem is that in  
NML terms a link does not have a port.   So we need to have a  
different name for the end points of connections.  I think the options  
are make up a NSI name for the end points, or to have NML define a name.

A couple issues with possible solutions.  a) If NSI define a name for  
endpoints then it will have to map back to NML concept for port in  
case of node and I am not sure what for a link.  b) if NML were to say  
that both links and nodes have ports, then node ports would not be  
physical, or perhaps they would be defined by a connector at the  
physical layer.

I note that in G.800 both links and nodes have ports, and where ports  
are connected there is a point.  This seems useful when saying that  
segments are connections between ports and segments are concatenated  
at points.  I am not sure how it fits with other NML concepts.

This seems a decision that needs input from NML to decide. Probably  
this will take some discussion and we may have to have interim names  
till we decide.

--
I would like to talk about these on the call tomorrow. We can continue  
online if needed.

  I will send a second email will a proposed outline of a architecture  
doc and call info.

Let me know if you have suggestions for other agenda items.

John