The MPLS WG Archive

["Angela Chiu" <alchiu@research.att.com>]

Mark,

There may be demands for both models. For example, an established carrier
who provides services to all kinds of clients may choose to have the control
of the optical network within the network while having clients signal
requests only. On the other hand, a relatively new carrier who owns the
transport network that provides only IP service may want a more integrated
control by having IP clients provide limited control to the optical network
besides signaling requests. In order to provide proper controls, certain
information of the optical network (not the full information) may need to be
propagated to the IP clients. So it would be desirable to design a set of
protocols that are extendable to support both models.

Regards,

Angela Chiu

-----Original Message-----
From: owner-mpls@UU.NET [mailto:owner-mpls@UU.NET]On Behalf Of Mark
Jones
Sent: Thursday, June 15, 2000 6:17 PM
To: clapp@research.telcordia.com
Cc: mpls@UU.NET
Subject: RE: RE: Information about optical routing in MPLS net.


Only time will be able to answer these questions.

Here's one perspective from a person who claims some expertise in
optical networking.  Hopefully, the industry will settle on a common
control plane to facilitate messaging between layers, but one layer is
unlikely to have full control of another.  Small homogeneous networks
may be the exceptions.  There are too many complexities in network
management of different layers to expect multiple layers to collapse
under one control plane (though the protocols in each may be the same).

The ODSI and OIF may define just such an interface, but unless carriers
accept it as a reasonable solution to management of their networks, it
won't be implemented on a large scale.  The ODSI and OIF have both been
anxious to show quick results, regardless of whether they met the
requirements of major customers.  I don't want to be too critical here,
because the ODSI and OIF output may be exactly what we need in our
networks, but the lack of rigor in development of their specifications
concerns me.  The OIF has jumped to solutions from the beginning,
without creating clear requirements for what network providers needed.
So, issues like whether carriers wanted an IP router to be able to
fully control an optical cross connect have been ignored.  (Sprint has
not participated in the ODSI as far as I know.)

The two camps on this issue, even within Sprint, fall into those who
love IP and those who only accept it as the primary growth protocol
today.  The IP lovers want IP routers to have flexibility to fully
control the optical network.  The other group is concerned by the
history of router failures, IP's limited and complex traffic
engineering, and the many other protocols that must be supported on a
common transport infrastructure.  The decision of whether IP routers
will have the ability to control an optical network may come down to a
political decision as much as an economic or technical one.

Mark Loyd Jones
Sprint
Technology Planning & Integration
913-534-5247
mark.jones@mail.sprint.com


-----Original Message-----
From: clapp [mailto:clapp@research.telcordia.com]
Sent: Thursday, June 15, 2000 2:25 PM
To: tli
Cc: clapp; mpls
Subject: FW: RE: Information about optical routing in MPLS net.


I agree with the notion of optical equipment reusing IP control
protocols
to manage the network, but to what extent will routers that are clients
of
an optical network manage that network?  The ODSI is defining an
interface
(using MPLS signaling protocols) to enable a router to manage an
optical
channel, and they say that ATM switches and SONET ADMs can do the same.

Managing a channel isn't the same thing as managing a network, but it's
a
beginning.  How far do people think it will go?

George Clapp
voice     973-829-4610
email     clapp@research.telcordia.com

At 11:28 AM 6/15/00 -0700, you wrote:

>  | I would support the view that having the router control the
optical domain
>  | is not good.  This comes down to a simple desire to see an optical
layer
>  | which is open to ALL clients - not just IP.  Of course it is
possible to
>  | allow the router to control the optical layer ... however, this
idea
>  | overlooks the fact that an operator may wish to carry many other
clients
>  | across an optical network.
>
>
>The optical plane clearly needs a control plane that provides a
(lambda)
>routing and signaling infrastructure.  This infrastructure is
fundamentally
>similar to the routing and signaling technology that we're already
using
>for MPLS.  Now, we can choose to re-invent this technology, or we can
>leverage the technology that we have already have developed and have
>started to deploy.
>
>Note that this argument is completely independent of the nature of the
>optical client.  Nothing here precludes non-IP usage of the optical
core.
>It would simply use the IP network as the out-of-band control plane.
>
>Regards,
>Tony