Metropolitan network evolution with dwdm-based transport
An optical transport network based on dwdm will address today?s bandwidth needs while enabling the transition to a data-optimized, multiservice infrastructure.
Brent Allen and James Rouse
Cambrian Systems Corp.
An optical transport network based on dense wavelength-division multiplexing (dwdm) technology is the next logical step in the evolution of the metropolitan network. This transition is critical to the success of the metropolitan service provider; the new infrastructure will increase revenues by supporting a broad array of new high-capacity data services. It also will reduce capital and operating costs through the elimination of service-specific network elements and the more efficient use of bandwidth.
Wavelength networking using dwdm technology will accommodate rapid bandwidth growth, provide unprecedented flexibility, and enable the transition of the service layer to data-optimized technologies such as Asynchronous Transfer Mode (atm) and Internet protocol (IP). The migration of all services to this new service layer, coupled with the expansion of the dwdm-based transport infrastructure, will lead to a simplified network by reducing the number of sublayers. The result will be a simpler, more cost-effective, higher-capacity, data-optimized network consisting of a multiservice infrastructure interconnected over a dwdm-based optical transport network.
The network of today can be modeled with three layers--service, infrastructure, and transport (see Fig. 1). The service infrastructure layer provides the functions necessary to provide end-user telecommunications services such as on-demand (dial-up or routed) connectivity. Historically, the technology used in service layer network elements has evolved as service types change. For example, the rapid growth in data services is causing explosive growth in the deployment of atm and IP switches and routers, whereas voice services continue to be provided by narrowband (64-kbit/sec) switches. Some services, such as leased lines, are fixed-bandwidth pipes provided directly by the transport layer, essentially bypassing the service infrastructure.
The transport infrastructure provides connectivity between the service elements over the physical media: fiber, radio, and copper. Fiber is a very high-capacity resource--much higher than is required for the typical connection between any two elements in the service layer. To exploit the capacity of fiber, transport layer network elements provide terminal multiplexing, add/drop multiplexing, and crossconnection so that many connections can share the same fiber. These functions are collectively referred to as bandwidth management.
It might be brought into question why the service layer network elements don`t provide the transport bandwidth management functionality and thereby eliminate the transport network elements. In a typical large office, over 50% of the bandwidth entering that office transits the office without needing to be processed by a service layer network element. Once again, this situation is driven by the physical locations of the offices, the physical locations of the service layer network elements, and the rights-of-way where the fiber is buried. A transport network element can transport a high-capacity trunk at a fraction of the cost of a service layer network element. For example, the cost of allowing 24 DS-3s (44.736 Mbits/sec) to pass through an OC-48 (2.5-Gbit/sec) add/drop multiplexer (adm) is much less than the cost of 48 DS-3 ports on an atm switch or IP router.
The transport infrastructure currently consists of sublayers (see Fig. 2). These sublayers are characterized by the granularity at which bandwidth is managed, and they chart a history of the growth and evolution of the transport infrastructure. The bandwidth management sublayer granularities shown in Figure 2 reflect the North American transport infrastructure. The transport infrastructure of other parts of the world can be similarly modeled, but the sublayer granularities would be different. There are three distinct sublayers in the transport infrastructure (not including the fiber) that have evolved over time, as the demand for bandwidth has increased in response to greater service demand and new services that consume more bandwidth have been introduced.
The 64-kbit/sec sublayer consists of DS-0 digital crossconnect systems (dcs 1/0) and other network elements that manage bandwidth at 64 kbits/sec and n ¥ 64-kbit/sec granularity. This sublayer is an artifact of voice-dominated service networks.
The 1.5-Mbit/sec sublayer was introduced because network growth made it impractical to manage all bandwidth at 64 kbits/sec, new services were being introduced at the higher rates, and an enabling technology (T1 and inline regenerators) made it possible for this rate to be carried over existing copper lines.
Similarly, the 50-Mbit/sec sublayer was introduced because network growth made it impractical to manage all bandwidth at 1.5 Mbits/sec, new services were being introduced at the higher rates, and an enabling technology (fiber optics) made it possible for this rate to be carried between service network elements.
Today, 50 Mbits/sec is no longer a big enough pipe for connecting many service network elements to each other. atm and IP switch vendors are already introducing trunk interfaces that accommodate rates in the gigabits-per-second range. Furthermore, growth in general is making it impractical to manage all bandwidth at 50 Mbits/sec.
Transport infrastructure evolution
dwdm is the enabling technology that will provide connections between the service layer elements at high speeds on the existing fiber plant and thus provide the next step in the evolution of the metropolitan transport infrastructure. In metropolitan networks, the primary requirement is flexible, high-capacity connectivity between multiple service layer entities located at various locations around the network. A dwdm-based optical transport network provides high capacity per fiber and high capacity per connection. Each dwdm wavelength provides a connection that can carry any protocol with a bit-rate ranging from 50 Mbits to 2.5 Gbits/sec and beyond. These wavelengths can be multiplexed with other wavelengths and added, dropped, and crossconnected at the optical level, eliminating the need to manage the bandwidth at a lower granularity when not required.
The transport network layering with the addition of a new dwdm-based layer is shown in Figure 3. This new layer provides:
The separation of wavelengths carrying different types of service (e.g., time-division multiplexing [tdm], atm, IP trunks, or high-capacity optical leased line), enabling a diversity of connections to the appropriate service elements.
The add/drop multiplexing of wavelengths to/from interoffice facilities to bypass the service layer network elements at intermediate offices.
Survivability at the optical layer in the event of network failures.
The multiplexing of wavelengths onto fiber-optic lines in preparation for transport at aggregate rates up to 80 Gbits/sec and beyond.
The new dwdm layer complements the finer-granularity layers by providing coarse-granularity bandwidth management. The finer-granularity elements, primarily based on Synchronous Optical Network (sonet) technology, provide bandwidth management, including multiplexing to 150 Mbits/sec and beyond. These rates are allocated to wavelengths in the dwdm layer.
The dwdm layer gives the transport infrastructure new capabilities not possible with tdm technology. A wavelength is not constrained by a fixed-rate time slot in a predefined multiplex protocol. It can carry any protocol--such as sonet, Escon, fddi, and Ethernet--and any bit rate, such as 150 Mbits/sec, 1.25 Gbits/sec, and 2.5 Gbits/sec. The decision about what protocol and bit rate to use for a connection between service layer network elements can now be based on costs and functional requirements rather than the constraints of the transport layer. Furthermore, the protocol and bit-rate carried on a given wavelength can be changed "on the fly" without altering the transport infrastructure, giving the network provider the ability to rapidly respond to service changes and growth.
dwdm is more cost-effective than tdm-based multiplexing when bandwidth greater than or equal to 150 Mbits/sec is required between service layer network elements (see Fig. 4). Note that Figure 4 does not include the total aggregate bandwidth carried on the fiber, which happens to be much greater in the case of dwdm. Instead, it depicts the granularity of bandwidth management at which dwdm technology becomes more cost-effective than tdm technology. If the value of fiber also is factored in, dwdm becomes even more economically compelling.
The increased cost-effectiveness of optical domain bandwidth management comes from the fact that optical channels that do not need to be terminated at a given location do not require back-to-back electro-optic conversions. For example, for a sonet OC-48 adm to drop a 150-Mbit/sec pipe, the adm requires two OC-48 terminations in addition to the 150-Mbit/sec interface. All traffic passing through the adm must go through the back-to-back OC-48 electro-optic conversion. However, a dwdm-based adm enables pass-through traffic to transit the node optically, thereby eliminating the intermediate electro-optic conversion and resulting in cost savings and increased networking functionality.
Network providers are deploying high-capacity, data-optimized technologies in the service layer to address the rapid growth of services. The two technologies at the forefront are atm and IP. As the network capacity consumed by data surpasses that of voice, it is anticipated that all services eventually will migrate onto the new multiservice infrastructures created by these technologies.
atm and IP network elements are interconnected with high-capacity trunks in the range of 150 Mbits/sec to 2.5 Gbits/sec and higher. The dwdm layer most cost-effectively manages these rates. Therefore, the finer granularity sublayers in the transport infrastructure will be phased out as the new service infrastructure expands. A much simpler network will emerge.
The network vision
Figure 5 illustrates the target metropolitan network, consisting of a single, data-optimized, multiservice infrastructure interconnected over a dwdm-based optical transport network. The evolution to this network vision will not happen overnight. However, a dwdm-based network provides the common enabling link between the architecture of today and the network of tomorrow.
The dwdm-based network provides a cost-effective, high-capacity, survivable, and flexible transport infrastructure in the future network. The elimination of multiple service network overlays and fine granularity transport network sublayers will reduce the number and types of network elements, creating a corresponding reduction in capital and operating costs for the metropolitan network provider. Transport interfaces that are specific to protocol and bit rate will no longer be required, which will reduce the network provider`s inventory costs while increasing network flexibility. At the same time, the data-optimized service infrastructure will enable increased revenues by supporting new high-value services and rapid growth.
High capacity is inherent in a dwdm-based solution. Each wavelength can support up to 2.5 Gbits/sec and beyond, while 32 or more such wavelengths can be multiplexed onto a single fiber. The resulting aggregate capacities of 80 Gbits/sec solve the network provider`s fiber exhaust issues while supporting high-capacity trunks between the service layer network elements.
A dwdm-based network provides the ability to route wavelengths with the same survivability capabilities as current sonet technology when deployed in a ring topology. In fact, the network can significantly enhance survivability by reducing the number of electro-optic devices in the network--a major source of equipment failures.
The flexibility of the dwdm-based network is derived from the protocol and bit-rate independence of the traffic-carrying wavelengths. Protocol and bit-rate independence is a key advantage of dwdm that enables optical transport networks to carry many different types of traffic over an optical channel regardless of the protocol (e.g., Fast/Gigabit Ethernet, atm, sonet, Escon, fddi, and video) or bit-rate (e.g., 100, 150, 240, or 622 Mbits/sec, and 1.25 or 2.5 Gbits/sec).
Protocol and bit-rate independence allows the network provider to make the best choice of service network element interface and rapidly respond to new service requests and the need for more bandwidth. For example, native data interfaces such as Gigabit Ethernet or native atm can be connected directly to the transport layer without costly adaptation. User requests for increased bandwidth or different protocols can be filled quickly, without the need to augment the supporting infrastructure. This method enables the network provider to realize increased revenue sooner and turn speed of service deployment into a competitive advantage. New services such as optical leased lines that provide end-to-end protocol and bit-rate-independent connections can be offered, attracting new revenue opportunities.
Optical transport networks based on dwdm technology maximize the use of the existing fiber plant while accommodating the exploding growth of new and existing services. They also enable the transition from multiple overlay service networks to a single, data-optimized, multiservice infrastructure. This new infrastructure will provide the connections needed to support a broad array of new high-capacity data services that will generate new revenue for the network provider. The new infrastructure also will reduce capital and operating costs through the elimination of service-specific network elements and through the more efficient use of bandwidth.
dwdm technology is poised for explosive growth in the metropolitan network environment. For metropolitan network carriers the evolution to dwdm-based optical transport networks will be a profitable journey toward the apex of network capacity, flexibility, and survivability. u
Brent Allen is product manager and James Rouse is market-development director at Cambrian Systems Corp. (Kanata, ON, Canada).