Re: Optical Link Interface

[Sudheer Dharanikota <sudheer@nayna.com>]

Hi:

My few cents .. being involved in these discussions.

Kireeti Kompella wrote:

>
>
> c) The model: master-slave vs. peer.
>
>    This is the key debate (in my opinion).  Is LMP-WDM a "natural
>    extension of LMP", as Andre says; or is the master-slave model
>    more appropriate in this context, as Osama says?
>

The issue here is who has the control over which resource.
The OXCs have control over the links and link bundles, and
the DWDMs have the control over the fault reporting behaviour,
the monitoring (may not be in the first release of the requirements)
and the mapping of logical nature of the links and link bundles
to the physical resources.

In my opinion there are and there will be DWDM and OXC
boxes which will not be capable of doing all the required
features (as mentioned in the above paragraph). This leads
to negotiation of features (or capabilities) between the two.

Hence I feel this protocol LMP-DWDM or NTIP should be a
peer-to-peer protocol.

>
> d) The transport protocol: TCP or raw over IP.
>
>    Does the reliability of using TCP as transport offer a real
>    advantage in this context?  Or do the complications of
>    managing failover with TCP outweigh this benefit?
>
>

The main goal of these protocols is to report a failure as
quickly as possible. A failure typically effect multiple
links. Another observation is that the two elements (DWDM and
OXC) need not be immediate neighbors. They can go through
DCN. But in either case the amount of traffic in the control
plane may not cause packet losses.

With these two observations I feel that TCP is an over kill.


Cheers,

sudheer