The MPLS WG Archive

[Shen Gangxiang <EGXShen@ntu.edu.sg>]

Hi, folks,

I have a query on the draft: Signaling requirements at the optical UNI,
draft-bala-mpls-optical-uni-signalling-00.txt

When a UNI-C sends a Light-path Create Request to a UNI-N, the message uses
(1) Source termination Point, IP address, (2) Destination termination Point,
IP address, (3) Source termination point, port, Channel, Sub-Channel, and
(4) Destination termination point, port, Channel, Sub-Channel to identify
the exact source and destination addresses of the lightpath. To achieve
this, this means that the UNI-C must have the information on the reachable
IP addresses of all OXCs (i.e. IP addresses that can be accessed by an OXC).
Based on this information, UNI-C decides the destination OXC and its logical
interface of the lightpath. If these address information could be exchanged
between UNI-C and UNI-N through "neighbor discovery" procedure, this
approach of course can work. 

However, I wonder if we could use the IP addresses of the initiating and
terminating UNI-Cs to substitute those information in (1)-(4). The detail
suggestion is as follows: using the "neighbor discovery" procedure, we
register the client IP address and user group identifier with the optical
network, then the optical network employs IGP protocol to flooding these
address-reachable information around the network; each time when an OXC
receive a Lightpath Create Request that includes the IP addresses of the
initiating and terminating UNI-Cs, it calculates the route for the lightpath
and then trigger the signaling process to build up it. After the light-path
is established successfully, the OXCs in the both terminations inform their
corresponding UNI-Cs on this event. 

With such a modification, the optical network will look like a black-box
except some interfaces to communicate with UNI-Cs. Moreover, there will be
less information exchanged on UNI interface. 

Above is my idea. If it is not so correct, please kindly point out. Thanks!

Gangxiang