Defining metro convergence and migration strategies for carriers
Metro Solutions / SPECIAL REPORTS
Equipment designers are moving bandwidth management out of the central-office bottleneck and onto the edge of the optical network.
The metropolitan public-network arena presents ample opportunities to equipment manufacturers and service providers for innovation. The issue is with metro convergence, where legacy protocols such as SONET/SDH, ATM, and frame relay are mixing with a new generation of optical technologies. That is further illustrated by optical aggregation and transport pushing closer to the edge of the network.
Despite the proclamations about "Internet Protocol everywhere," no single protocol will dominate. Carriers have too much money invested in existing infrastructures to write them off anytime soon. SONET and SDH are not going away. Internet Protocol (IP) and Gigabit Ethernet "over glass" are definitely coming, if not already here in some areas. But carriers are not looking for some overlay network to magically drop from the sky and solve their problems with routing and switching increasingly heavy, data-laden traffic streams through networks mostly constructed for voice.
Instead, carriers are demanding systems that will help them make a smooth and cost-efficient transition to the evolving network paradigm. One way equipment designers are addressing the problem is by moving bandwidth management out of the "central-office bottleneck," which creates unnecessary back-haul and transport between the central office (CO) and traffic aggregation points around the network and moving it to the edge.
The industry is rapidly moving toward a model of "edge intelligence" and platforms designed expressly for the metro area that support Layers 0-3 of the open systems interconnection (OSI) model are emerging. These platforms integrate all of these layers into a single network element: DWDM-fueled capacity enhancements at Layer 0, add/drop multiplexer (ADM) and digital-crossconnect functions at Layer 1, ATM switching at Layer 2, and IP routing at Layer 3.
This integrated intelligence now resides, or soon will, at the optical edge. To boost efficiency, such platforms integrate easily with existing time-division multiplexing (TDM) networks. Therefore, carriers can transport existing TDM traffic and have a seamless migration option to ATM (any port to any port) and IP routing (Border Gateway Protocol-4 (BGP4), intermediate system to intermediate system (IS-IS), open shortest path first (OSPF), Multiprotocol Label Switching (MPLS), etc.).
The ability to easily migrate from 100% TDM to 100% ATM to 100% IP, or any mix thereof, allows carriers to accommodate their TDM and ATM networks today and, when the time is right, add some IP services or migrate to an all-IP network without changing any customer or network interface hardware. Switching TDM, ATM, and IP natively and offloading this switching circuitry onto optional fabrics is the key to an efficient protocol-agile solution (see Figure 1).
When evaluating these emerging service platforms, carriers are learning that they have to pay attention to "the slots" before expecting to hit the jackpot. That means this new class of equipment has to provide a much greater range of "universal slots" (or multiprotocol interface ports) than are available to date. This protocol agility means, for instance, unchannelized DS-3 (45-Mbit/sec) ports must be upgraded via software to channelized DS-3 packets over DS-3 and ATM over DS-3. This feature saves money in truck rolls and equipment swaps when upgrading services.
Today, multiple protocols are present in the metro-access network, as they will be in the future. Both voice and data must be picked up from business parks and residential areas and switched at the edge in one integrated platform. Traffic that needs to travel to the core network must be statistically multiplexed to more efficiently utilize the bandwidth between the edge and the core.
At local central offices (COs), traffic from multiple rings must be groomed and switched between the rings in a single network element. In regional COs, the ability to segregate these statistically multiplexed, or "dirty" DS-3s, and groom them back into clean DS-3s for handoff in the CO, is a necessity (see Figure 2). Protocol-agile bandwidth managers are the product of choice to perform these functions in the metro-access network.
This design represents the evolution of the metro network to a third-generation "protocol-agile" model. Carriers need such agility because they face unpredictable traffic mixtures and service churn. This unpredictability has spawned the emergence of "x over wavelength" equipment that allows carriers to transport SONET, Gigabit Ethernet, ATM, frame relay, and more, as long as they have an "optical grade" signal of at least OC-3.
However, edge traffic is not running at this speed and carriers need help handling the flood of protocols. Port utilization in a multiprotocol environment is very low with existing equipment be cause the number of available slots is limited and the ports on any given module are locked in to certain protocols. That wastes resources and makes the already difficult process of trying to predict the required mix of protocols and ports even more onerous.
"Fill rate" analyses are used to illustrate the advantage of the third-generation multiprotocol port systems over equipment introduced only a year or two ago. If a carrier wants to add clear-channel DS-3s to a bandwidth manager, a working and protection DS-3 card must be added (see Figure 3). If the carrier then wants to add DS-3s with trans-multiplexing capability to the bandwidth manager, a flexible architecture would allow them to be added to available ports on the same DS-3 card. Architectures that are not as flexible would have to add more DS-3 cards to accommodate the trans-multiplexing function, and in many cases, system capacity is sacrificed when trans-multiplexing functions are added.
If the carrier then wants to add ATM user-to-network-interface (UNI) DS-3s to the network element, a protocol-agile bandwidth manager accommodates them on available ports of the same DS-3 card, while other bandwidth managers must add more ATM-specific DS-3 cards. The same scenario is repeated when adding clear-channel and ATM signals at the OC-3 rate-the protocol-agile bandwidth manager adds fewer circuit packs than the others. It becomes apparent that protocol agility translates into much lower first-in costs for the carriers. Figure 3 also shows that protocol-agile interfaces allow for more efficient utilization of physical space in the network element, leaving more room for growth of different and new services than other bandwidth managers. Further analyses show savings in terms of rack space in the central office, lower equipment costs, and faster provisioning.
One application for the highly integrated platform of a bandwidth manager is a ring collector (see Figure 4). In a ring collector, several access and interoffice rings bring traffic to a CO, and the traffic from these rings is aggregated and groomed between the rings via a crossconnect. Since a bandwidth manager can handle multiple rings in one network element and provide the crossconnect functions of grooming and switching between the rings internally, multiple network elements (and multiple racks of equipment) can be replaced with just one shelf of equipment (see Figure 5).
Consider an example involving DSL services and DSL access multiplexer (DSLAM) aggregation. DSLAMs are ATM-based products, making them perfect candidates for the third-generation solution for efficient deployment of DSL services across a metro region. A third-generation product can aggregate traffic from multiple DSLAMs across a SONET ring. Each DSLAM supports up to 1,472 asymmetric DSL lines in this scenario. A take rate beyond this figure would require multiple DSLAMs connected via an ATM switch. Instead, the new platform aggregates all the DSLAM traffic to a single ATM switch port via a highly efficient data pipe across the metro ring.
The service provider can now expand the benefits of its already flexible DSL platform, enabling the operator to deploy DSL from very small take rates to high ones in a cost-effective manner, without a network overhaul or the risk of premature technological obsolescence. The carrier can now implement oversubscription, which translates into additional revenue and market share.
This example also addresses the issue of a low initial deployment cost. Regardless of the take rate or traffic patterns of end users, the operator can increase the customer base simply through added port concentration and the flexible statistical distribution of bandwidth. Depending on the actual network traffic patterns, an end user may be offered additional bandwidth at reduced cost. Furthermore, third-generation metro platforms featuring ATM backbones enable carriers to offer improved quality of service (QoS) in the form of more granular services to end users, enabling the customer to pay for exact services when and where they are needed.
When fully configured for crossconnect and add/drop capacity, one next-generation metro shelf replaces four racks of broadband crossconnects, two racks of wideband crossconnects, and one full rack of SONET ADMs. With collocation real estate prices at a premium, next-generation leading edge port density provides a major space-saving benefit to operators.
Various equipment vendors seem to offer similar value propositions. Upon examination, however, those similarities lie among the first- and second-generation products, which have done quite well. Still, network requirements have evolved so quickly that third-generation products now offer the best answers for moving intelligence to the edge of the metro optical network.
Dana Hartgraves is vice president of marketing at Metro Optix Inc. (Plano, TX). She can be reached at www.metro-optix.com.