[Nsi-wg] notes from call 8/12

[John Vollbrecht]

On the call were Jeroen, Tomohiro, Joan, Chin, Freek, Guy, Gigi and  
myself.

Discussion

1. Joan describe work he is doing with describing NS agent  
interfaces.  He is working on slide set and will hopefully be able to  
present it next week or the week after.  Some discussion
	- discussed the difference between dataplane which can be modeled by  
G.800 and the control plane.  Tomohiro said that we should have a  
third plane that describes coordination between NS agents differently  
than coordination of dataplane [I may have this wrong - Tomohiro might  
want to comment}.
- Chin talked about architecture document being worked on for DOE that  
describes a service plane, and as I understand it the service plane  
describes all interactions with NS agents.  Interactions between  
agents is (I think) always with web services.   Chin will see if it is  
possible to share this doc with the NSI wg.
- there was some discussion of how Lookup Service and Path Computation  
capabilities fit into this.   These seem like Information service  
similar to DNS or BGP for IP.
- This is a subject to be discussed further in the group - by email  
and on calls.

2. Guy is working on a context section for document and will have a  
draft for next week

3. We discussed  G.800 and NML and NSI.
Consensus was that G.800 was a good way to describe NSI type  
networks.  Concepts of G.800 map nicely to NSI concepts.  Guy suggests  
that we consider using G.800 naming for NSI.   I am not sure that  
works completely, but would be good if it could.  One issue is that G. 
800 concept of subnetwork is NSI concept of network.  There may be  
other issues.  We should consider this.

We discussed how G.800 and NML interact.  Seems that G.800 describes  
concepts that are included in NML schema.  This is mostly done, NML  
group is working on details of schema now.  Once schema is done then  
it is represented in a language.

Discussed the difference between NML and GMPLS description of  
adaptations.  My take is that GMPLS defines specific adaptations in  
detail, while NML wants to have abstract adaptations with capabilities  
assigned to each instance.

Gigi raised the question of whether G.800 had been implemented by any  
vendors or providers.  Jeroen agreed to ask the same question on  
Friday on the NML call.  I am wondering the same for GMPLS.
-

We will have a call next week - Wed at 9ET.

please add, correct or comment on notes -

John