Hello all-
I have been working on description of transport plane for NSI. I
think what is coming out of this makes the coordination simpler in
some ways and also raises a problem to be resolved.
1) NSI is basically a resource reservation and instantiation
protocol. The resource it deals with is connections between ports. A
connectionin NSI map to a links in NML quite nicely. So this is the
part that is simpler- there is no need to distinguish between networks
and links as NSI has been doing.
2) Resources controlled by an NS provider agent are a set of ports
between which it can provide transport connection. This can be
represented by a group in NML. The group of NML resources is a
combination of nodes and links. The problem is that the NSI
resource may a terminate with a link or a node. If it terminates with
a node, the node has a port. If it terminates with at link, links
currently have no port. This is the problem - NSI resource [NML
group] has ports on the edge while NML does not define port links, and
link may be last topological element in the group.
To resolve this problem NSI could call end points something other than
ports, or NML could define links as having ports. I am not sure which
way is better. Having ports be specific to nodes makes it possible to
describe switching and termination capabilities for the port, which
don't exist on links.
Having ports exist on links seems to correspond to the G.800 and ASON
view where nodes and links have ports and where ports connect to each
other is a point. This also makes the names on the edge the same for
both NML and NSI which might be desirable.
Comments or suggestions?
John
