Make it simple

Feb 27, 2002 12:00 PM

Today's metropolitan networks are in a state of technical and economic flux. Functioning as the critical bridge between the access and long-haul network segments, metro networks are faced with the challenge of having to adapt to changing requirements of carriers: "build to demand, not to forecast." This paradigm shift has exposed inefficiencies and complexities associated with current metro backbones, which are composed of complex overlays of optical transport rings, digital cross-connect systems (DCSs) and other switching equipment.

Despite the availability of many new metro equipment solutions today, elements of the traditional metro transport infrastructure have not evolved for many years. In fact, today's metro core hub sites comprise numerous elements that occupy large footprints and are power-intensive, difficult to manage and non-scalable. The impact of this inefficiency becomes clear with the realization that up to 70% of a carrier's overall costs are in the metro network, with most of the cost being operational expenditure.

Today's market dynamics compel carriers to seek a rapid migration of metro networks to more simplified, flexible and scalable architectures, to enable them to respond to market demands more rapidly, efficiently and, above all, more cost-effectively.  The term 'migration' is key, as carriers stress their desire to evolve existing legacy networks using non-disruptive technologies, in contrast to building more overlay networks using disruptive technologies.

Challenges in today's metro networks

Metro networks were originally built to support to TDM-only voice traffic. With the explosion in data services, carriers must now accommodate and manage networks that are more than 50% data, and because of this changing mix, they must also accommodate a shift in the predominant origin and destination of metro traffic. In voice-centric networks, 80% of traffic typically originates and terminates in the same local metro network. Not so for data-centric networks, where 80% if the traffic is destined for the long-haul network.

To date, carriers have responded to changing traffic types and patterns by deploying more equipment: overlays of Sonet ADM- and DWDM-based optical transport equipment, DCS equipment for traffic grooming and Layer 2/3 devices for supporting ATM and IP/MPLS-based services. Network scalability has become a high priority, and the typical optical line rates have jumped from OC-3 (155 Mb/s) just a few years ago, to OC-48 (2.5 Gb/s) and OC-192 (10 Gb/s) predominating today.

In particular, DCSs have taken center stage as a key scalability problem and cost burden. While needed to provide the critical bandwidth grooming function in the backbone, DCSs are expensive on a per-port basis, usually occupy large footprints (several rack bays per network element) and are managed as stand-alone non-intelligent systems that do not scale very well. A metro core hub site (Figure 1) will typical comprise discrete wideband (WB-DCS) and broadband (B-DCS) cross-connect systems to support the bandwidth management requirements for VT1.5/DS1 and STS-N circuits, respectively.

Among the biggest challenges carriers continue to face is whether to over-provision the network in response to forecasts, or to build "just in time." Because data is non-deterministic, data-centric (IP) networks are more difficult to accurately forecast than voice networks. Therefore, should carriers invest to over-provision their networks, and then later realize that demand was much lower than forecasted? Or, should they build to serve a lower, more stable level of core capacity and scramble to rapidly deliver to the unpredictable occasion demand spikes? Given responsive architectures, carriers can accomplish the latter with less risk than ever before.

New generation metro network

Migrating today's metro networks to a more simplified, converged architecture helps carriers respond to market demands more rapidly and cost-effectively. In addition to slashing infrastructure expenses, a metro network that is substantially more software-driven than in the past will enable carriers to migrate in a modular and non-disruptive fashion from the legacy base, while allowing the introduction and rapid provisioning of new broadband services such as gigabit Ethernet, storage area network services and high-speed VPNs.

New-generation optical transport switches (OTS) are being developed by several vendors to leverage expertise and functionality in the carrier-class switching, routing and signaling domains. The OTS variants combine optical transport, grooming and switching with bandwidth granularities that extend from wideband (VT1.5/DS-1) level to super-broadband (STS-N) and intelligent provisioning capabilities. Such a new network element is also intended to support a diverse range of traffic types (TDM, ATM, IP/(G)MPLS). Some are also scalable to multi-terabits capacity without service disruption.

While the OTS class of network element is most typically being designed as an all-optical networking platform, carriers are showing a preference (particularly incumbents with large CO, tandem and super-POP offices) for highly granular bandwidth grooming down to VT1.5 (DS-1) levels to support the massive T-1-base, which would otherwise lack a cost-effective on-ramp to the long-distance network. Despite the market fixation with broadband service opportunities, T-1 and T-3 private lines continue to grow steadily and represent a huge proportion of carriers' revenue streams. For this reason, the OTS class system would form an integral part of the new generation metro network, bridging the gap between the legacy TDM-centric network and the increasingly more data-centric architecture. Such systems should also be capable of operating with existing ring-based networks while supporting mesh-based topologies as required. Again, this flexibility allows carriers to leverage their existing base while eliminating network overlays.

Carriers as varied as ILECs, utilities-based carriers, data carriers and IXCs indicate that they see the value of the OTS as highest, near-term, when deployed at large metro hub sites. These are some of the most complex network locations today, where traffic must be terminated and groomed before passing onto the long-haul network or other carrier's networks. In North America, a nationwide carrier typically has as many as 70 to 75 of these types of metro core hub facilities.

Network architecture migration is clearly not a simple undertaking. Carriers must carefully balance capital and operational investments against revenue and margin considerations, while also addressing competitive forces and new market opportunities. Network strategies based on marginal improvements are insufficient--carriers looking to introduce new products into their network must consider the incremental costs associated with network integration, OSS integration, testing and interoperability. For a successful business case, carriers require step-function economic and operational improvements, which meet immediate objectives of optimizing their operational model while migrating to a new-generation architecture that opens new avenues for profitability. The realities of risk management and cash flow/budget constraints also dictate a network migration strategy that is set out in phases. Often the initial phase of migration entails simplification of the operator's ADM and DCS transport network (Figure 2), typically the most expensive and operationally intensive part of the metro backbone network.

Once the optical transport network has been collapsed into a single-layered OTS-based architecture, the carrier can plan and implement a second phase of simplification--the introduction of multiservice intelligence into the new-generation architecture. Today, carriers typically use ATM core switches exclusively for cell grooming and routers exclusively for packet grooming at metro core hub sites. While necessary at network access service points, these service-switching platforms are prohibitively expensive if used solely as multiservice bandwidth managers. If the OTS has been designed with cell and packet bandwidth management intelligence, carriers have the opportunity to cut over this functionality to the OTS and extend the efficiency of the network to full multiservice switching convergence. Figure 3 shows the new-generation metro architecture following second phase of migration.

New provisioning trends

The speed with which services can be created and provisioned has become vital as carriers compress their time-to-revenue from new offerings. As noted previously, most services travel substantially over TDM network segments. Thus, increasing service velocity for TDM networks becomes even more important. TDM-based networks are inherently more rigid than data networks, and they employ hop-by-hop manual provisioning. This old paradigm does not support high-service velocity--consequently, the new generation metro should ideally feature point-and-click end-to-end provisioning, irrespective of service type.

Generalized multi-protocol label switching (GMPLS), a new protocol suite based on MPLS, has garnered serious attention among carriers for its ability to automate traffic engineering and end-to-end provisioning through a common control plane unifying the optical and IP layers in the network. GMPLS takes MPLS features such as label-switched paths and extends them to optical switching media, a scope not originally encompassed in MPLS. GMPLS includes not only packet switching, but also TDM, wavelength (lambdas) and spatial-switching. Its common control plane seeks to manage complexity, simplify traffic-engineering problems and dramatically reduce network bottlenecks. This protocol promises to automate network-resource management and provisioning for all service types, and provide the basis for an open and interoperable next-generation optical network.

What carriers ultimately require is a network architecture comprising multi-vendor end-to-end "point and click" provisioning for all traffic (Figure 4). With GMPLS, metro access products--such as MSPPs and ADMs--can signal each other and the OTS in the metro core to provision connections. In turn, the OTS can signal long-haul transport systems to provision the long-distance connections. Using products that comply with the Optical Interworking Forum's (OIF) UNI 1.0 specification, carriers have a solution to migrate their networks toward an open, end-to-end architecture that can deliver service-agnostic provisioning times of seconds and minutes, rather than days and weeks.

While the metro network is in the midst of great flux, it has also become an area of tremendous excitement. Even as this part of the network presents some of the biggest challenges for carriers, it also presents the some of the biggest opportunities for growth and deployment of new-generation technologies, addressable by solutions that will deliver a step-function improvement in bandwidth efficiency, network simplification, service diversity and provisioning. The migration of the metro network to a flatter, simpler and more flexible network will give carriers the ability to respond and adapt to changing demands without expensive network rebuilds.

Using optical transport switches as key building blocks for the new-generation metro backbone network, carriers can implement a network architecture that combines optical transport, bandwidth grooming and switching into a single layer. This architecture would be multiservice-intelligent and support rapid provisioning of all service types. Carriers will be able to respond faster and more efficiently to changing and variable demands: scale the network cost-effectively, build to demand, and software-enable new services for increased revenue opportunities.

Sab Gosal is Director of Product Marketing for Polaris Networks.

Visit Polaris Networks online.