Tài liệu The optical last mile doc

· · · · · · ·
VIRTUAL ROUNDTABLE
Networld+Interop, Atlanta, Fall 2001
Where will it lead us?
last mile
The
To assess the issues and opportunities optical technologies and networks bring
to end-users, we invited six leading experts to our semiannual roundtable dis-
cussion, scheduled for Sept. 11, during the Networld+Interop show in Atlanta.
Unfortunately, that give-and-take session had to be cancelled. In its place, we
created a virtual roundtable on “The Optical Last Mile,” with our original panel
of experts participating via e-mail, resulting in this interesting and informative
dialogue on the optical revolution.
Is the optical last mile really necessary?
Why or why not?
Sandra McWilliams: Fiber is the only available me-
dium that achieves the optimal bandwidth, reliability, flexibil-
ity and security required by today’s enterprises. As the way
we conduct business contin-
ues to evolve, new video and
data applications will tax the
limits of existing infrastruc-
ture. Advancing the optical last
mile is a proactive, pragmatic
approach to the imminent
growth of bandwidth demand.
Mark Fishburn: Sep-
tember 11th obviously mag-
nified many market trends
already in place, but may
accelerate business benefits

for the high-speed access market. Many pieces to the
puzzle—increased demand for home entertainment to
upcoming technology and metro-focused service provid-
ers—make delivery of that fiber economical. Layoffs
will presumably lead to an increase in SOHOs looking
for low-cost, high-speed connectivity.
Cüneyt Özveren: Given a cyclical economy like this
one, it becomes very difficult to scale IT departments up
and down as market conditions change. A solution for this
problem is outsourcing IT requirements. Optical last mile
enables a high-bandwidth access to enterprise networks,
which, coupled with quality of service, enables application
service providers to offer outsourced IT services.
David Montanaro: Long term, there is little doubt
that the bandwidth requirements of business and residential
users will surpass the ability of any other media to handle
it. If fiber is the end game,
why waste time and money
on interim solutions now?
Brendan Mullooly:
The optical last mile is nec-
essary in the overall
scheme; however, we see
it as an ethernet-based, end-
to-end solution. Only fiber
provides virtually limitless
bandwidth, as opposed to
fixed-wireless or copper-
based solutions. Where en-
terprise users are within easy reach of the fiber-optic infra-

structure, they can benefit from virtually unlimited band-
width of optical ethernet. Offerings from traditional telcos
and ATM-based networks are unable to match the robust
capabilities of optical ethernet in terms of flexibility,
scalability, cost, bandwidth and speed of deployment.
Lawrence Prior III: Customers are demanding the band-
width. Optical technologies are the only way to bridge the gap
between the need for bandwidth at the network edge and the
Sandra McWilliams, director, fiber cable
management, fiber-optic division, ADC:
“The benefits garnered by optical
technologies are virtually limitless if its
infrastructure is designed to work with
multiple technologies and grow as
demands grow.”
OPTICAL
OPTICAL
What are the challenges and concerns for
enterprises considering optical technologies?
Mark Fishburn, vice president—business
strategy, Spirent Communications, and a
member of the board of directors, 10
Gigabit Ethernet Alliance: “September
11th obviously magnified many market
trends already in place, but may
accelerate business benefits for the
high-speed access market.”
Cüneyt Özveren, founder, chairman of
the board and CEO, Atoga Systems:
“Given a cyclical economy like the one

we are going through, it becomes very
difficult to scale the IT departments up
and down as the market conditions
change. A solution for this problem is for
the enterprises to outsource their IT
requirements.”
huge optical bandwidth capacity already in existence at the
network core. As more bandwidth-hungry applications are de-
veloped, the bottleneck between the edge and core will become
more acute. For example, end-users with 11-Mbps wireless
LANs are not going to be satisfied with only a 1-Mbps
network connection—especially when efficient high-capacity
fiber-optic networks are just a few hundred meters away.
Mullooly: Oftentimes, there is a perception that opti-
cal-networking projects are a bigger task than they actually
are. Deploying optical technologies, such as optical ethernet,
allows enterprise network managers to add computing band-
width where it makes the most sense, easily add users to
the network and reduce the need for specially trained IT
staff to manage the network. The primary benefit from
optical ethernet is that it increases the available bandwidth,
while relatively reducing the cost of WAN connections—
plus enabling the use of familiar and inexpensive LAN
technologies and protocols, such as VLAN.
Enterprise network users and LAN managers may be un-
comfortable with fiber and all its ramifications, such as mul-
timode, single-mode, connector types and distance require-
ments. At the enterprise level, there is no need to worry
because the service provider will connect the enterprise to the
fiber network, providing the necessary demarcation point be-

tween the LAN and the MAN, usually as simple as deploying
a media converter between the two. As a result, enterprise
users do not see any obvious, discernible difference because
the change from one to the other is totally transparent, with
ethernet running on both sides of the connection.
Prior: Enterprise customers have to develop optical
networks that are flexible and scalable as the technolo-
gies improve. The near-term challenge will be—figura-
tively and literally—to close the loop, to finish optical
networks and bring benefits of optical technologies to
end-users. Business custom-
ers want efficiency, reliabil-
ity and low costs. As de-
mand for bandwidth in-
creases, as more complicated
applications come on the
market, businesses will have
to adopt new services in or-
der to stay competitive. As
competition grows and bud-
gets contract, there will be
even more emphasis on find-
ing ways to connect enterprise customers to the thou-
sands of miles of fiber already at the network core.
McWilliams: Two scenarios exist: renovations of existing
infrastructure and new construction installations. The decision
between deploying optical technologies or tapping existing
copper infrastructure hinges on how each solution meets cur-
rent and future requirements, network compatibility, and cost.
New construction lends itself to the deployment of fiber. The

investment provides a vastly superior long-term solution, offer-
ing maximum bandwidth capability and minimal ongoing ex-
pense. Once fiber is deployed, enterprises must then choose
optical technologies that best suit network requirements.
Özveren: Enterprises have traditionally relied on very
high levels of interoperability. Optical technologies have
not reached that level yet. Enterprises planning to use
optical technologies in their networks now should be will-
ing to commit to one vendor once they make their choice.
Fishburn: As a supplier of performance analysis and
service assurance solutions, we are connected to all the
individual competing technologies—DSL, cable modem,
PoS, ethernet to 10GbE, wireless technologies and the
applications that run over them—MPLS, voice over IP,
QoS, Internet applications. Certainly, bringing these new
products to market faster gives a sense of optimism that
the technology can be deployed in a reasonable time frame.
Montanaro: One view is “build it and the applications
will come.” The huge pipes that fiber could support would
create dramatic differences in the way we do business. Imagine
what a real-time, full motion, HDTV-quality conference system
for meetings and education could do for your business. This is
perhaps even more compelling after recent events.
McWilliams: The business climate is changing. Many busi-
nesses are cutting operating
costs. Videoconferencing,
shared data files and online
applications are part of the new
business paradigm. The trend
toward a dispersion of employ-

ees, decrease in travel and re-
liance on online applications
has opened the door for inno-
vative technologies that allow
leveraged assets to produce best
value results. The benefits gar-
nered by optical technologies are virtually limitless if the infra-
structure is designed to work with multiple technologies and
grow as demands grow. Look for DWDMs to allocate specific
wavelengths for individual customers; test access to slice main-
tenance costs and preserve network integrity; and optimal pro-
tection, scalability and reliability to finally reside at the desktop.
Özveren: Coupled with quality-of-service guarantees, band-
width allows enterprise applications to be outsourced
completely. This is a win-win scenario for carriers and enter-
What can be done with the bandwidth
that optical will make available?
· · · · · · ·
VIRTUAL ROUNDTABLE
Networld+Interop, Atlanta, Fall 2001
prises, since—on one hand—enterprises get to spend money
incrementally as opposed to investing in large IT infrastruc-
tures. They can adjust the “volume” of services they purchase
as their needs change. This is also a win for the carriers, since
it increases the revenue stream they get from their customers.
Fishburn: The extension of high-speed core network-
ing, provisioned by metro providers, will be one-half of
the delivery model for data transport. The other is the
extension of ethernet-over-fiber services to the last mile.
The ethernet in the first-mile initiative—it’s the last mile

for the service providers, but the first mile for the user—
had its inaugural meeting as the IEEE P802.3ah task force
in October. The 802.3ae committee has focused on extend-
ing high-speed networking at ethernet costs to the 10-Gbps
MAN/core network. Now, 802.3ah will focus on low-cost
delivery of high-speed ethernet to the office and residence.
Mullooly: Today, most applications do not use all the
available bandwidth delivered by fiber. Most companies
find that one T-1 line at 1.544 Mbps is ample for their
needs and, if more is needed, additional T-1s are added.
The difference is substantial in comparing the cost of a T-
1 with a 10-Mbps optical drop at a location. While the use
of 100 Mbps or 1000Base-LX is rare for typical enterprise
users, it is suitable for applications where substantially
more bandwidth is required, such as video-on-demand,
high-end multimedia, remote storage networks and vertical
applications, like telemedicine, healthcare and printing.
Prior: The obvious applications involve the conver-
gence of voice, video and data. The one constant in every
new business application is a need for greater bandwidth.
What’s less obvious is the mission of bridging the explod-
ing bandwidth demands of the LAN with the capacity of
the network. Certainly, the recent tragedies brought into
focus teleconferencing, storage area networks and disaster
recovery—all bandwidth dependent—and many businesses
now see them as essential.
Fishburn: Bringing the low-cost ethernet business model
to the metro network will be a key enabling link from the
edge of the network to the overprovisioned core. Key
factors will be price and availability—and will ethernet in

the first mile be deliverable anytime soon.
McWilliams: Look for proven technology from a proven
company. Today’s turbulent financial environment is unfor-
giving; vendors with exceptional records of quality service
must be chosen. Free-space optics (FSO) is a promising, but
relatively unproven, technology. Poor propagation conditions,
physical obstruction, and security will always pose realistic
threats to the integrity of the free-space signal. Because of
these uncertainties, land-line backup systems will need to be
available, negating any positives gained with the technology.
Prior: Experience is invaluable. Look for suppliers who
have real customers in the market. Ask for references. Is it real
or “in development”? The best test is to see if the supplier can
drop ship equipment today. Too many suppliers claim general
availability, then keep you waiting for two quarters. The market
will sort this out in a very Darwinian way. I would also look
for suppliers with strategic partners who can bring expertise
and experience to the table beyond their own company’s prod-
uct. With free-space optics, if a business moves, the equipment
can be redeployed, making FSO a much more efficient long-
term solution than other technologies.
Montanaro: Surveys show that customers value com-
pany performance and technology in the purchase of network
gear. Most new customers are interested to find out who else
uses your gear; therefore, existing customer relationships are
very important. Regarding free-space optics, we’d recom-
mend FSO where short turn-up time and high bandwidth are
most important. For cases in which higher reliability is most
important, we would stick with a land line.
Özveren: Most important is reliability. Since optical tech-

nologies are relatively less mature, certain technology choices
by some vendors may be less reliable than enterprises are
used to. The second important factor is plug-and-play capa-
bilities. A lot of the optical technologies require hand tuning
power levels after the equipment is installed, and any changes
typically require retuning the network. New technologies
are becoming available that make DWDM plug and play.

Montanaro: All of the above, though I’d include inertia,
also. In the U.S., we have over 190 million access lines. A
ball that big takes time and a lot of effort in the three areas
identified to start moving. I’m afraid that progress sometimes
takes longer than the media or the technologists have the
patience for within the access arena. DSL is a good example.
Özveren: Technology is not a barrier, since some vendors
have shown that this can be done with the equipment that
they have been shipping. Economics is typically not a barrier
either, since a converged device which incorporates multiple
devices will typically cost less than the combined cost of all
those devices, and result in less operational expenses, since it
is much easier to manage one box. The barrier is really
politics. Enterprises and carriers have developed separate or-
ganizations internally for different layers of the network. A
converged solution requires buy in from all these in order to
be accepted into the network. Faced with financial challenges,
most enterprises and carriers are reorganizing to remove mul-
tiple layers. Once these layers are removed, it will be much
David Montanaro, CEO/chairman of the
board, NEC Eluminant Technologies Inc.:
“One view is ‘build it and the applica-

tions will come.’ The huge pipes that
fiber could support would create
dramatic differences in the way we do
business. Imagine what a real-time, full
motion, HDTV-quality conference system
for meetings and education could do for
your business.”
What should be considered when choosing
an optical supplier? When deciding between
hard-line connections vs. free-space optics?
· · · · · · ·
VIRTUAL ROUNDTABLE
Networld+Interop, Atlanta, Fall 2001
What is preventing converged networks:
technology, politics, economics?
Most large U.S. cities have one or more
optical metro networks. Will more networks
be built in these Tier 1 cities? When can
most Tier 2 and 3 cities expect optical ser-
vice? How will the growth of these networks
impact costs to the end-user?
and currently, the unsteady market and a reduction of
competitors has allowed service providers to remain cau-
tious about deploying optical metro networks.
Montanaro: You would be hard pressed to find a U.S.
town of more than 25,000 that did not have some type of
fiber deployed in the ground. Chances are it belongs to the
local ILEC and is being used to serve digital loop carriers,
or maybe a large business. Trouble is, there may not be
enough businesses to attract competitors for the ILEC.

Mullooly: Supply and demand will rule the prolifera-
tion of networks. In cases where fiber is not already in-
stalled, the upfront cost to lay these networks is exorbitant.
In most Tier 1 cities, a fiber glut exists. Less expensive,
faster and more flexible than traditional telephone networks,
optical ethernet provides the bandwidth and the distance that
most users desire. Where the fiber network resides close to
the enterprise network, fiber to the business makes more
sense, with service providers using media converters to
serve as the optical access point to connect the LAN to the
MAN. In some Tier 2 and 3 cities, fiber networks are being
deployed today because of specific demand and an existing
infrastructure to support the network. Most Tier 2 and 3
cities do not contain existing networks, which reduces the
need to deal with network integration issues. Service pro-
viders such as EtherLECs are investigating the possibility of
building their optical metro networks in these cities that will
provide service at a competitive price. In those cases where
the use of fiber is cost-prohibitive, it makes more sense to
deploy microwave or wireless for the last mile.
Prior: The rollout of optical networks is continuing, despite
the much-publicized slowdown in the market. History has
shown that short-term economic forces won’t stop the natural
progression and adoption of technologies that really make
sense. With more competition and advances in long-haul tech-
nology, more small cities will soon receive optical capabilities.
The problem will be finding cost-efficient and fast ways to get
that bandwidth to end-users and to complete networks where it
will not be economically feasible to lay fiber. That’s where
free-space optics comes in. If the networks can reach them,

end-users will be the biggest beneficiary of this optical growth.
Not only will they get the bandwidth they crave, the services
will become increasingly affordable.
Lawrence Prior III, CFO, LightPointe:
“History has shown that short-term
economic forces won’t stop the
natural progression and adoption of
technologies that really make
sense.”
Brendan Mullooly, director of
European operations, IMC Networks:
“Oftentimes, there is a perception that
optical-networking projects are a
bigger task than they actually are . . .
At the enterprise level, there is no
need to worry because the service
provider will connect the enterprise to
the fiber network.”
· · · · · · ·
VIRTUAL ROUNDTABLE
Networld+Interop, Atlanta, Fall 2001
easier to remove the layers in the network, as well.
Prior: Conflicting standards, self-interests and cutbacks
in budgets have certainly slowed the rollout of all-in-one
networks. But convergence is happening. Recently, in
Manhattan, we experienced firsthand that the demand is
there. One of our partners actually gave up its own VoIP
equipment to help out a customer who was looking to
restore its network using FSO links.
Mullooly: With converged networks, technology definitely

plays a role because it is difficult to integrate a variety of
different protocols all into one network. Coupled with the
complexity of different vendors’ respective hardware and soft-
ware solutions, it becomes an expensive and time-consuming
proposition to achieve true convergence. Optical ethernet elimi-
nates many of the challenges because of the support of the
standard by many vendors and their solutions.
The economic landscape is changing with the advance-
ment of metropolitan optical ethernet networks. The pro-
liferation of ethernet service providers in Europe and the
U.S. has further broken down this barrier. These ethernet
service providers, EtherLECs, have built their networks
from scratch by renting fiber from other providers using
a simple architecture. The proposition of EtherLECs is to
provide 10 Mbps or 100 Mbps at the same price of a
typical T-1. As a result, EtherLECs build lower-cost net-
works that make efficient use of the MAN bandwidth.
The provisioning of optical ethernet connectivity to corpo-
rate customers is now well established. The speeds offered—
10, 100 or even gigabit—can be utilized by larger customers
for hosting multiple servers, SANs or even internal voice-
over-IP traffic. Applications that can utilize the bandwidth
provided by optical connectivity are well established.
McWilliams: Their deployment depends heavily on the
well-known theory of supply vs. demand. Service provid-
ers will hesitate to build metro networks until demand for
optics intensifies and competition poses a significant chal-
lenge. When this happens, end-users will reap the ben-
efits and avoid much of the costs. It is a waiting game,
Reprinted from

Communications News
, November 2001
Copyright © 2001 by Nelson Publishing Inc. • www.comnews.com