RE: O-UNI/GMPLS node implementing CR-LDP

["Morgan, Peter (Engineering)" <ptmorgan@emea.att.com>]

Perhaps this will a configurable option on the "intelligent" box. Much the
same way as a BGP speaker knows whether  to run IBGP or EBGP.

Regards,
Peter.



-----Original Message-----
From: Remi Theillaud [mailto:remi.theillaud@atosorigin.com]
Sent: 15 January 2002 11:41
To: oif-signal@oiforum.com; ccamp@ops.ietf.org
Subject: O-UNI/GMPLS node implementing CR-LDP


I have one question regarding the extended discovery mechanism for nodes
that host both a GMPLS application and an O-UNI application (I'm not aware
of any standard forbidding such a configuration).

My current understanding:

_ For GMPLS, a node determines whether it plays the active role or the
passive role for a LDP session by comparing its own source (or transport) IP
address with the source (or transport) IP address of the LDP peer.
_ For OIF O-UNI, the O-UNI client always plays the active role; the O-UNI NE
always plays the passive role.

Assuming that the LDP extending discovery mechanism is used. If a node
implements the CR-LDP protocol for, let's say, both GMPLS and O-UNI NE, how
is it supposed to behave when it receives a LDP Targeted Hello message ? I
mean: which rules should it use to determine whether it should play the
active or passive role for the LDP session ?

Now assuming the LDP basic discovery mechanism. Same question. Maybe here
the control channel from which Link Hello messages are received is either
used for O-UNI signaling, or for GMPLS signaling, not both, so the rules
that should be used can be inferred from the control channel; makes any
sense ?

Remi Theillaud
Atos Origin