Cost-effective Service Delivery Will Make Metro
DWDM Real
7/28/99 DWDM systems viable for metro networks must not
only provide capacity, but also aggregate a variety of service
delivery and bandwidth configurations.
By: Doug Green, Chromatis Networks
Contents
Lagging behind long haul
Unpredictable growth
Integrated solutions
Carriers have deployed dense wavelength division multiplexing
(DWDM) systems extensively in long haul networks, driven by a
critical need for bandwidth brought about by the explosion in Internet
and data services. The economic case for these deployments is
clear: Capacity-expanding alternatives such as new fiber builds and
parallel networks can cost many times more than DWDM solutions.
Internet and data services also generate capacity demand in
regional networks, and carriers have begun to deploy DWDM
technologies in short-haul, point-to-point applications. But the
business case for these applications is not obvious as it is for long
distance networks, because synchronous optical network (SONET)
upgrades and additional fiber installations are not as
cost-prohibitive.
Even with the market for short-haul, point-to-point DWDM
applications in its infancy, vendors also offer ring-based DWDM
products for metropolitan area networks. While most carriers
express interest in these systems, most industry experts believe that
large-scale deployment of these systems is years away.
top
Lagging behind long-haul
The reason that DWDM was readily accepted for long haul, but not
so quickly for metropolitan rings lies in the differences between the
issues faced in long haul versus metro networks. DWDM satisfies
straightforward capacity needs in long-haul networks, but
inadequately addresses the complexity of service delivery in metro
networks (see Figure 1). To address the needs of the metro
network, new integrated optical multiplexing technology must format
bandwidth for a variety of services and aggregate and groom traffic
from those services, in addition to scaling optical transport capacity
and providing protection switching.
In the long haul network, the issue is simply bandwidth. Traffic
reaches the network backbone groomed and aggregated into full
OC-48 or OC-192 signals, which directly map into DWDM
wavelengths. From the DWDM perspective, there is not much
difference between carrying voice, data, or a mixture of the two.
They are simply bits in SONET-framed signals. DWDM aptly
provides bandwidth expansion or fiber relief by accommodating
multiple channels for these signals per fiber.
While DWDM provides a quick fix to the long distance network
problem, it does not solve all the problems facing the metro network.
The goal of the metropolitan area network is to deliver a variety of
services in a variety of bandwidths, which presents a more complex
problem than bandwidth expansion. In other words, DWDM must
address a different set of issues to gain acceptance in metro
applications.
Figure 1: Service delivery issues characterize metropolitan network needs.
top
Unpredictable growth
Data traffic growth is an essential issue for the metro network, but it
does not share all of the same characteristics of the long haul traffic
growth that sparked DWDM acceptance. Long haul traffic consists
of the combination of hundreds or thousands of end users' data
traffic into DWDM-caliber signals, and therefore always justifies
DWDM deployment. As long as the DWDM system can add new
wavelengths, it should be able to handle traffic growth. In metro
traffic, however, less aggregation has occurred. Bandwidth demand
at a particular site could arise from steady consumption of normal
voice and T1-level data services, or a web hosting business could
move in and demand an OC-12 at a single site.
DWDM rings handle unpredictable growth, but only after the traffic
reaches a certain capacity. Components for the technology are too
expensive to be economically feasible for delivering anything less
than an OC-12 to OC-48 per site. Once the traffic at a site reaches
this 'clip level,' DWDM rings allow network operators to provision
wavelengths quickly in response to high bandwidth growth. The
problem is that few networks in the near future will need that much
bandwidth at every site, or even at a majority of the sites. WDM's
inability to economically accommodate low bandwidth sites at the
same time it provides high bandwidth for applications that need it is
one key inhibitor to near term metro DWDM acceptance (see Figure
2).
Figure 2: Metro DWDM scales up in bandwidth economically for high-bandwidth
applications (left), but the cost of optics makes DWDM too expensive for sites with
lower bandwidth requirements (right).
Another issue facing today's metropolitan networks is that several
separate networks are needed to deliver services. For example, a
SONET network provides transport, but a carrier may build an
asynchronous transfer mode (ATM) network or an Internet protocol
(IP) network on top of SONET to deliver services. Multiple networks
mean more boxes, more network management systems, and more
complex provisioning. Building a DWDM network under the SONET
network to provide more transport capacity only makes the situation
worse, adding a new network layer and suite of network elements to
pay for, manage, and provision (see Figure 3).
Figure 3: DWDM rings scale capacity, but also prompt more complexity with
additional layers of equipment needed to fill the extra bandwidth.
DWDM ring proponents suggest that services should run "natively"
over DWDM optics with no underlying SONET infrastructure. This
sounds good in theory, but as mentioned before, services must
present wavelength-class bandwidth demand to justify the cost of
DWDM optics at every node. While devices such as gigabit routers
can run directly over DWDM, other devices that deliver services
within the metro network, including DSLAMs and low line rate
routers and ATM switches, cannot run directly on DWDM. This
incompatibility requires another network layer to convert raw DWDM
channel capacity into useable form.
top
Integrated solutions
The poor job of reducing service delivery complexity that DWDM
rings perform today contribute to their slow acceptance in metro
networks. To gain rapid acceptance of DWDM technology in the
metro network, system designers must shift priority from capacity
relief with long haul economics to address network complexity and
service delivery faced by metropolitan networks today and in the
near future. A new class of integrated optical multiplexing (IOM)
systems is emerging to accomplish this task.
IOM products integrate multiple technologies to address both the
complexity and scaling issues of metro networks in an integrated
fashion. The approach provides many benefits aside from the raw
bandwidth scale provided by DWDM rings, including the following:
The ability to manage multiple layers of the network as a single
homogenous network, allowing faster turn-up of new and
existing services and lower network operations cost;
Small footprint to allow the deployment of scaleable
transmission systems in small collocation spaces for
competitive local exchange carrier (CLEC) applications;
Direct connection to network elements that deliver services,
such as DSLAMs, cable modem headends, ATM switches,
and IP routers, eliminating the need for intermediate network
elements to map these devices into wavelengths; and
Grooming and aggregation functions that span multiple
services and network protocols.
Figure 4: An integrated optical multiplexing system provides the ideal metropolitan
network element architecture.
In the ideal IOM architecture (see Figure 4), the optical subsystem
provides access to transmission services and protection. The
multiplexing system performs aggregation and grooming of traffic
from multiple services, including both data and voice traffic. This
could be accomplished either with an ATM fabric or a combination
of ATM and time division multiplexing (TDM), or perhaps with IP as
the technology evolves to carry voice. The service subsystem
supplies tributary interfaces to connect directly to legacy voice
services and emerging network elements, such as DSLAMs,
routers, and ATM switches. Common management spans all
subsystems and reduces the complexity of service provisioning.
Technology integration overcomes one of the hurdles for DWDM in
the metro by reducing, rather than increasing, the complexity of
service deployment. However, requirements for DWDM-scale
bandwidth on a site-by-site basis, not a ring at a time, presents a
problem. Integrated optical multiplexers must also succeed
economically at lower bandwidths, while providing DWDM-scale
bandwidth quickly and on a site-by-site basis.
One solution to providing either high or low bandwidth scale per site
is to design a ring that maps services onto both a 1310 nm shared
ring and onto dedicated DWDM wavelengths on a selective basis
(see Figure 5). In Figure 5, network element 1 is accessing both the
shared 1310 channel and a DWDM wavelength, while element 2 is
only accessing the shared 1310 channel. Element 3 accesses all
wavelengths and the shared channel, and also has an integrated
function for switching traffic between wavelengths.
Figure 5: Metro DWDM design lets operators determine bandwidth scale on site by
site basis.
The network design presents three ways to connect network
elements:
1.Direct connectivity over the inexpensive shared
channel—where no DWDM capability is required;
2.Direct connectivity over dedicated wavelengths between
elements; and
3.Bring DWDM wavelengths and/or the 1310 channel into a hub
with an integrated switching function located in one or more of
the network elements, which reduces the wavelength
requirements for full connectivity between nodes.
The three configurations enable network operators to deploy DWDM
when and where necessary and maintain full connectivity between all
network elements.
One additional requirement is key. As capacity requirements grow
at each site, each network element must upgrade independently,
whether it be a migration from shared ring to DWDM capability or an
addition of more wavelengths. Upgrades must be done in-service,
with no impact to sites that do not require additional bandwidth.
There is a case for DWDM in the metro, but not in a 'pure
technology' form. The same bandwidth solution that worked well in
the bandwidth-oriented long haul does not translate directly to the
service-oriented metro network. DWDM plays a critical role in
providing bandwidth scale, on a selective basis, within the new
generation of integrated products. Integrated optical multiplexers will
gain acceptance not as technology solutions, but because they
provide what the carriers need: Faster time-to-market for new
services, reduced complexity of provisioning, and lower cost of
operation for both new and existing services.
top
About the author…
Doug Green is vice president of marketing with Chromatis
Networks, PO Box 30450, Bethesda, MD 20824-0450, e-mail:
doug@chromatis.com, web address www.chromatis.com.
www2.fiberopticsonline.com{D7E5A791-42CE-11D3-9A54-00A0C9C83AFB}&Bucket=HomeFeaturedArticles
|
