The metropolitan core: fertile ground for the chosen few
As service providers identify critical network requirements, competing vendors are facing off to prove which architectures best meet those requirements-and only the strong will survive.
ARUN BELLARY, Metro-Optix
The convergence of voice and data communications is happening now, in the core of metropolitan fiber networks. In the past 12 months, millions of dollars of venture capital have poured into private optical-networking equipment companies that are targeting the metropolitan core with a powerful combination of new technologies. As operators evaluate and begin deploying these next-generation technologies, it will become increasingly apparent that certain product architectures are better suited than others for metropolitan applications. The main differentiating features like scalability, flexibility, legacy network compatibility, and seamless migration to target network topology will decide who will break from the pack and run for successful deployment.
The next few years will bring dramatic changes and challenges to the communications industry. Pressure will occur on several fronts: exponential bandwidth demands, introduction of new services, operator preferences for "one-stop" shopping, reduced service-activation times, evolving standards, and a growing demand for improved reliability. To thrive in this environment, service providers need futureproof networks that maximize scalability and flexibility, while simultaneously supporting a wide range of existing, emerging, and new technologies. While the new networks must be cost-effective and reliable, the most daunting challenge will be the need for them to evolve from each service provider's unique legacy network.
Keeping pace with Internet traffic growth is the greatest challenge. Since overtaking voice traffic as the dominant protocol just two years ago, Internet traffic has continued its explosive growth. By the end of this year, it will have grown as much as four times larger than voice traffic, and with bandwidth demands doubling every eight to 10 months, Internet traffic could easily top voice traffic by 40 to 50 times within five years.
This traffic surge is acutely felt in access and metropolitan networks where increasing bandwidth demands are straining existing facilities. These capacity issues can only worsen as access speeds in excess of 1 Mbit/sec penetrate more of the consumer market and business traffic expands to 45 Mbits/sec and beyond. But the growing pains are not limited to capacity. Internet Protocol (IP) addresses are becoming scarce. Operators are continuously deploying new services and demanding the network evolve to meet the new requirements. Businesses are demanding better Internet quality and high availability of services as greater amounts of their revenues are Internet-based.
The gradual amalgamation of legacy carrier and enterprise data networks has led to expectations of future multimedia services. Voice, imaging, video, and data must all be delivered and managed on a single carrier-class platform.
New networks must adapt easily and quickly to these changes while supporting legacy services-without sacrificing quality and availability or compromising essential operations like lifeline services. Voice, video, and data traffic must seamlessly coexist as voice and data communications continue to converge. Existing single-technology fixed structure (time-division multiplexing-TDM) and single-wavelength network topologies are quickly becoming obsolete, because they do not offer a homogeneous way to carry these services.
New-generation equipment will provide effective solutions to these issues and enable service providers to evolve their networks to handle these challenges. The new generation of network equipment will differentiate itself by its scalability, density, flexibility, adaptability, availability, and compatibility.
The explosion in Internet traffic is forcing the network to scale to handle more and more traffic. New-generation network equipment will be forced to exhibit unprecedented ability to not only match delivery capacities to the applications, but also expand the capacity without interrupting service. Switching capacity will scale to match interface rates, and every effort will be made to utilize facilities to their maximum. Support for DWDM and statistical multiplexing of data traffic is quickly becoming an absolute requirement. DWDM multiplies the capacity of the fiber facilities, while statistical multiplexing eliminates bandwidth waste by dramatically improving utilization rates.
In the metropolitan arena, OC-3 (155-Mbit/sec) and OC-12 (622-Mbit/sec) add/drop multiplexers (ADMs) are now commodities. Speeds from OC-12 to OC-48 (2.5 Gbits/sec) will be common.
Best-in-class systems will support OC-192 (10 Gbits/sec) with narrowband optics for seamless integration with DWDM equipment. In addition to legacy TDM transport, new-generation metropolitan equipment will support data traffic natively. Best-in-class equipment will also support full statistical multiplexing for all data traffic, whether cell- or packet-based.
While capacity and service demands increase exponentially, available floor space for new equipment remains relatively constant. Floor space is often a limiting factor. This problem is most pronounced for service providers utilizing collocation areas where space is leased by the square foot and competition for space is fierce. Operators must continue to increase capacity per square foot.
For metropolitan applications, new-generation equipment has raised the OC-48 line-side density bar to four protected line-side interfaces with 192 protected DS-3 (45-Mbit/sec) drop interfaces in a single 7-ft rack. Best-in-class equipment will support up to 12 protected line-side interfaces and 288 protected DS-3s, without reducing capacity if other drop-side interfaces are selected.
Products like City-Stream from Metro-Optix can support densities of up to 192 OC-3s or 48 OC-12 ports with over 30,000 T1s (1.5-Mbit/sec) or 3,000+ STS-1 (52-Mbit/sec) switching capacities at the physical layer (PHY) alone. There is also the ability to convert these "agnostic" PHY ports into ATM user-to-network interfaces (UNIs) and packet-over-SONET (PoS) interfaces through simple software provisioning.
Today, access and transport network elements only carry circuits of a fixed size. This drawback restricts the types of services offered by TDM networks. Newer services like LAN interconnection do not easily fit into standard TDM circuits at DS-1 (1.544-Mbit/sec), DS-3, E1 (2.048-Mbit/sec), or E3 (34-Mbit/sec) rates, thereby stranding bandwidth or constraining the flow of traffic with circuits that are too small. The lack of crossconnection and statistical multiplexing close to the network edge wastes bandwidth and delivers inferior services to business users.
The new generation of network equipment attacks this paradigm in two ways. First, it eliminates the bandwidth bottleneck at access points to the public-network backbone by supporting wider varieties of interfaces. Gigabit Ethernet, Fast Ethernet, OC-192/OC-48/OC-12/OC-3, DS-3, electrical STS-1, PoS, and ATM UNI are just a few examples. Second, they enable a variety of traffic types to be transported over single transmission media. The ability to support multiple protocols and technologies is critical for the network services provider to seamlessly migrate today's TDM network to the next- generation data-oriented optical network, while still providing traditional TDM transport. Ideally, network equipment can be optimized for 100% TDM traffic, 100% IP traffic, 100% ATM traffic or any combination of these protocols, without penalizing applications where one or more of these technologies are not deployed. The service provider needs the capability to change these configurations at any time without sending service personnel to the field to rearrange network interfaces or change circuit packs.
Competing equipment providers will differentiate from each other based on switching capacities and interface capabilities. Features such as full IP routing versus packet concentration, best effort versus quality of service (QoS), label switching, protocol support, and breadth of PHY interfaces are differentiators. An interesting fallout from having a wide selection of interfaces is the need for equipment to provide as much granularity as possible to allow the service provider latitude in picking and choosing as many different types of interfaces in the same equipment as possible. Best-in-class equipment maximizes this flexibility.
For metropolitan applications, new-generation equipment supports multiple switching technologies in a single platform. All this equipment supports STS-based TDM switching, and most of it supports virtual-terminal (VT) switching. Other switching technologies that may be supported are ATM, frame relay, packet concentration, and packet routing.
For network interfaces, OC-3, OC-12, OC-48, DS-3, DS-1, and electrical STS-1 are commonly supported. OC-192, narrowband optics, Gigabit Ethernet, and Fast Ethernet interfaces may also be supported. Some equipment provides protocol-independent interfaces, while others are protocol-specific and require the deployment of a physical interface based on the protocol contained on the line. For instance, an M13 interface circuit pack cannot support an ATM UNI DS-3.
Best-in-class equipment delivers fully scalable switching of 100% TDM (VT/STS), 100% ATM, 100% IP, or incremental combinations of all-up to the aggregate capacity of the line interface rates. Full statistical multiplexing is supported at Layers 2 (ATM) and 3 (IP) to enable large quantities of ATM virtual connections or IP flows with virtually unlimited bandwidth granularity.
Today's networks contain several different types of network elements: one to provide an end-user service, a second to multiplex, a third to switch, and a fourth to transport the services. Building networks with these elements is expensive, and such networks are difficult to operate. Future technologies could exasperate this situation. To overcome these challenges, future networks will be composed primarily of multilayer bandwidth managers that support legacy TDM-based services as well as packet- and cell-based services. This new network architecture should support any proportion of services, whether based on TDM, ATM, or IP.
Service providers can flexibly provision any type of service to run over a homogeneous transport infrastructure and change the service mix at any time. Bandwidth efficiency has improved, and switching and routing resources are better utilized by distributing service creation to the edges of the network, thus reducing costs for the service provider and end user. Best-in-class bandwidth managers will combine protocol-independent transport with protocol-dependent switching and processing resources to enable support for not only today's technologies, but also future technologies. This flexible bandwidth manager architecture futureproofs the service providers' optical networks by providing a seamless migration path from legacy services to packet- and cell-based data services.
Metropolitan-based bandwidth managers provide support for TDM and one or more additional technologies. Best-in-class bandwidth managers support TDM switching, Multiprotocol Label Switching (MPLS), ATM switching, and IP routing through protocol-independent transport to optional protocol-dependent switching fabrics. The protocol-independent internal transport and protocol-independent interfaces enable future technologies to be easily supported through new switching fabrics.
Voice service providers are long accustomed to having to provide highly available lifeline-quality services. Fresh in the minds of many of these providers are congressional inquiries into service outages that affected large portions of the long-distance network. Up to now, data service providers have been largely immune from similar inquires. With businesses now linking revenue to Internet services and commercials playing to that theme, it is reasonable to assume that higher and higher levels of availability will be demanded from data service providers.
New-generation equipment will provide highly available services through the generous use of internal equipment protection via redundant circuit packs and external facility protection switching through one of many standardized automatic protection switching (APS) systems (linear 1+1, linear 1:1, UPSR, BLSR). Most protection switching is completed within 50 microsec of failure detection. The ability to provide a combination of physical and logical meshes, using DWDM, SONET or ATM, is another arsenal for the service provider to improve the robustness of the network.
Best-in-class metropolitan equipment supports 1:1 equipment protection switching for synchronization, control and switching fabrics (including IP), and 1:n (0 <n < 8) DS-3 / STS-1 interface protection. A full range of APS options are supported, including linear 1+1, linear 1:1, unidirectional path-switched rings (UPSRs), and 2- and 4-fiber bidirectional line-switched rings (BLSRs). APS protection is also available for ATM virtual-path rings, IP rings, and protection bundling, and ring interconnection is provided for multiple rings.
Networks centered around multilayer bandwidth managers put additional demands on fully compliant network interfaces, protocol operation, and control operating systems. Compliance must be provided for Layers 1 (physical, SONET, and synchronous), 2 (ATM) and 3 (IP). Compliance must be guaranteed for all switching fabrics, network synchronization, National Equipment Building Standard physical requirements, central-office footprints, and power inputs. Best-in-class equipment also supports worldwide operation-both the American National Standards Institute and European Telecommunication Standards Institute marketplaces.
Today's legacy networks support multiple types of services with fragmented overlay solutions (see Figure 1). Network service providers must deploy and maintain one network element for each OSI (open-systems-interconnection) layer (e.g., physical, data-link, network layers). Typically, these network elements are provided by multiple equipment vendors and require separate element management systems and support organizations.
New-generation equipment simplifies the management of the network by integrating the functions of several different legacy products into a single cost-effective platform that is the foundation of homogeneous optical networks (see Figure 2). These products can include traditional access and transport ADMs, digital crossconnect systems, frame relay access devices, ATM service access multiplexers, ATM edge switches, and IP routers.
New-generation metropolitan-based bandwidth managers collapse two or more of these functions into a single manageable entity. Best-in-class metropolitan-based bandwidth managers collapse all of these functions into a single entity with a common unified control-plane interface.
In addition to capital-equipment savings, consolidating multiple network elements into one can dramatically reduce time-to-revenue for operators and simplify the provisioning process. Provisioning times are reduced when a single network element is involved in activating the service and providing the bandwidth to carry that service. Intelligence in the unit reserves the required bandwidth based on the specified bandwidth requirements and class-of-service.
The challenge is to integrate the element with operation support systems (OSSs) that cross the transport and data domains. The easiest way to accomplish this feat is to speak the same language as the OSS does today. As network-management systems evolve, a common interface will emerge to provision a new virtual circuit or IP route with a single point-and-click. With the right solution, this development translates into profitability for carriers.
The metropolitan-area market is quickly changing. The competitive tapestry consists of more than a dozen new and established companies, touting everything from next-generation ADMs to full-featured bandwidth managers. In any crowded field like this, changes are inevitable. Some will survive, some will thrive, some will be consolidated, and some will abandon the race.
Arun Bellary is chairman and CEO at Metro-Optix (Santa Clara, CA). Prior to founding Metro-Optix, he was vice president and general manager of Ericsson's broadband access products business. For more information, visit the company's Website at www.metro-optix.com.