What's next for Nortel?
VP Brian McFadden discusses the company's stake in optical networking.
By Annie Lindstrom
Nortel Networks' late 1998 purchase of Cambrian
Systems Corp. (Kanata, Ont.) completes a circle the
Brampton, Ont.-based corporation began with its acquisition
of Bay Networks Inc. (Billerica, Mass.) last year.
Nortel, a leading synchronous optical network (Sonet) vendor, decided to buy,
rather than build, its metro dense wavelength division multiplexing (DWDM)
system. The decision signifies that data networking at higher-than-Sonet speeds
in the metro area could be a market reality sooner, rather than later.
Brian McFadden, vice president and general manager of the company's optical
networks applications unit, has assumed the task for making Cambrian's
OPTera Metro DWDM system a part of the Nortel product portfolio.
Cambrian designed the OPTera specifically to accept and transmit
non-Sonet-based traffic reliably, cost-effectively and — most important of all
— profitably.
While Nortel continues to build and ship OC-192 (10 Gbps) Sonet systems at
its St. Laurent-Quebec plant, McFadden works just a stone's throw away from
the manufacturing floor, planning the company's quick entree into the metro
DWDM data networking world. We visited with McFadden in January.
AN: Five years from now, how will optical networking analysts
look back on 1998 and 1999?
McFadden: What you'll see in the next five years is that the things you've seen
in the backbone or the core network you'll start to see in access and metro
types of topologies. That's why we bought Cambrian. The way that people
build networks in the metro area is going to be transformed into an optical
networking scenario.
We'll be in the business of manipulating photons. You can call it wavelength
switching, optical add/drop … but we'll be directing light in the optical domain
to reduce interconnection costs and bandwidth scalability costs as we are
required to by our customers.
If everyone in New York City went out and bought a [Nortel] 1 Meg Modem
tomorrow and hooked it up to their access networks, we'd have a big problem.
So we've got to solve that. If we are going to have high-definition TV, digital
radio and wireless access to the Internet, all of those bits are going to end up on
fiber someplace.
The fiber is going to be aggregated closer to the subscriber than it's been in the
past. So you are going to have to have this cost-effective fiber connectivity
technology called metro DWDM.
What is the biggest optical networking challenge going to be?
I think it's going to be optical network OA&M [operations, administration and
management]. No one will deploy an optical network if they can't solve those
issues. It goes beyond interfacing to [Bellcore operations support systems]. It
will encompass everything that we have built into the intelligence of Sonet and
finding a way to replicate that in the optical domain.
AN: Why did you decide to build a standalone interface for
your long-haul DWDM systems?
McFadden: What's happening involves the customers' service boxes.
Traditionally, on the customer premises you would have a single box — let's
call it an OC-3 multiplexer [mux] — which would aggregate all of the services
onto a laser and transport the traffic back to the central office [CO]. What you
are seeing is that the service boxes on the customer's premises don't have to be
an OC-3 mux anymore. They could be an ATM [asynchronous transfer mode]
switch, an IP [Internet Protocol] router, a gigabit Ethernet switch/router or a
video codec [coder/decoder] with a fiber interface on it.
As those service interfaces migrate into the optical domain, it makes sense to
have a DWDM system that is open to any kind of optical service interface that
you can think of.
We have optical interfaces on our DMS [CO] switches now, and on our
routers and ATM switches. Three or four years ago, those were all DS1 and
DS3 interfaces. Even frame relay switches have optical interfaces. So as that
proliferates, and it's still a multivendor world — where customers want to buy
an ATM switch from one vendor, an IP router from another vendor and a
Sonet mux from another vendor — then it makes sense to have a DWDM open
interface going forward.
AN: Are incumbent optical networking vendors letting
technical wizardry stand in the way of moving into all
optical-networking faster?
McFadden: In greenfield networks, there are a lot of things that don't really
need to be in the way. One issue [for incumbent carriers] is all sorts of fiber that
is already installed that have given characteristics. No one is going to go around
and replace it all, so you have an installed base issue in some places.
You also have to make your network backward-compatible with services that
you have today. You can't say, "You have a DS1 service, so I can't carry you
on my network anymore." It's fine to say, "If everything were IP, this is how
we'd build a network," but how are you going to handle all the legacy stuff
that's out there? You can't just say to a carrier, "Don't worry about your legacy
stuff."
There is lots of talk about being able to transmit a signal across the country or
the world optically. Is that going to happen soon?
People think that some day they may be able to do it, and they are talking about
solitons and other things. But right now most long-haul fiber spans are in the
400-to 600-mile range, and at the end of each span you have to regenerate,
retime and reshape each digital pulse.
AN: Let's understand what we are trying to transport here. We
are not transporting photons; we are trying to transport data
bits. The question is, what's the most cost-effective way to do
that?
McFadden: Right now, even though you could say it would be more
cost-effective to do it optically, you can't do it. The characteristics of fiber
transmission are that if you go beyond 600 miles, you've got to come back into
the electrical realm and recreate the pulse.
But didn't erbium-doped fiber amplifiers [EDFAs] revolutionize the distance
you could go along a fiber span not so long ago? Isn't there some product or
technology in the future that will change the realities of what you just said?
I don't think there's one component that will change it. There are people that
say it could be done by solitons, which are a way of generating a pulse; but it's
not clear that you could get 320 Gbps of solitons onto the same fiber. So if you
had to put in 320 fibers to get a soliton that would go 10,000 km, that would be
a different cost equation. You'd have more fiber and less electronics. There's
always a tradeoff between fiber and electronics.
It's a difficult equation and not one you can solve with one piece of equipment.
You have to look at the network. Nortel has a close relationship with Corning
Inc. [Corning, N.Y.]; one of the things we look at in those discussions is the
tradeoffs between fiber and electronics. If we could solve this all in the fiber
domain, we'd do it.
AN: Is the reason you want to sell the enterprise, access and long-haul
pieces of a network because a multivendor network will be too
complicated?
McFadden: All of the pieces we are putting together stand on their own right.
For instance, the Cambrian business stood on its own. They were in a position
to execute on their own.
Some say they weren't, and that they weren't making money.
They were a startup. How many start-ups make money in the first year?
Was their business plan really executable?
We wouldn't have bought them if it wasn't an executable plan. What we've
done is taken the risk out of the execution. Part of the issue they had was sales
people on the streets. Now they have a large, educated sales force. And I think
that's part of the equation. But did they have a viable product and business
plan? Absolutely.
AN: Are you going to keep the Cambrian name?
McFadden: We are still working through the branding. We want our brand to
be Nortel Networks — the operative word is ‘network' with an ‘s.' What that
means is, no matter what kind of network you want to build and under what
kind of criteria — including open and multivendor — we can help you build it.
Building unified networks is what we are trying to do. By unified we mean that
there should not be a difference in any service set that rides on the network. If
you can make a telephone call in three seconds, why should you have to wait a
minute-and-a-half for a Web page? If you can call an operator and get any
telephone in the world in a minute, why do you have to spend an hour on a
search engine?
AN: What do metro carriers need to know as they dive into
DWDM?
McFadden: First, they have to get comfortable with the technology. There is a
lot of concern about their fiber plant's ability to accept DWDM. There aren't
any big problems, just concerns: What if I have lots of mechanical splices?
What if my fiber is over 10 years old? I think that all of those can be worked
through. Engineers were put in this world for one reason, and that's to engineer
around problems.
They have to understand that optical networking is not as great a leap from
Sonet networking as some would think. We have to work with them on how to
get their craftspeople comfortable with deployment and comfortable with the
technology and the products, just the way the [interexchange carriers] have
become familiar with it over the last couple of years.
|
