To remain competitive, carriers need a cost-effective and flexible optical-switching technology that leverages SONET's potential to deliver innovative services.
Rafat Pirzada
Cyras Systems Inc.
The information contained in the terabits of complex, mixed-traffic traversing Synchronous Optical Network/ Synchronous Digital Hierarchy (SONET/SDH) networks is transforming commerce and society. For metropolitan-area carriers, however, it's a different kind of phenomenon-one reaching epidemic proportions. Existing metropolitan area networks (MANs) cannot adequately handle today's traffic, especially the increasing data.
It's only going to get worse. Traffic in MANs is becoming increasingly delay- and time-sensitive. Circuit-switching equipment, specifically designed to handle voice and data over point-to-point, time-division multiplexed (TDM) connections, was never intended to process and transport these complex payloads. Compounding the problem, industry research indicates that between now and 2005, a major demographic shift in MAN transport will occur.
During the next five years, rapid adoption of the packet-switching technologies used to carry data traffic is expected in MANs. Asynchronous Transfer Mode (ATM), frame relay, and native Internet-protocol (IP) services will supplant circuit switching and become the dominant protocols used to carry voice and data traffic. These protocols will transport 80% of the total traffic at optical line rates up to OC-192 (10 Gbits/sec), according to market researcher RHK (South San Francisco). That will resolve the majority of traffic time and delay issues. Unfortunately, 20% of the remaining network traffic will need to be supported by aging circuit-switched TDM equipment.
Despite all the hoopla, native IP will account for less than 50% of the total packet-traffic volume because class-of-service (CoS) and quality-of-service (QoS) capabilities are still virtually nonexistent during "native" IP transport. Thus, most carriers will not run mission-critical applications over IP. To incorporate CoS and QoS features, IP has to be augmented through sophisticated middleware such as Enron's.
The technology shifts and the legacy support requirements place MAN carriers in a difficult position at every interconnection point along the network-whether at the core, on the edge, or in the local loop. Incumbent carriers' entrenched TDM infrastructure must still be supported, as these operators gear up to meet the surging demand for packet traffic. Customers will inflict even more pain during this transition. Sophisticated end users will want flexible bandwidth options, provisioning for multiple circuit and packet protocols, better transport efficiency, and predefinable QoS levels.Figure 1. Trans-metro optical-switching technology combines the functions of multiple SONET platforms, offering scalability from DS-1 (1.544 Mbits/sec) to OC-768 (40 Gbits/sec), full-function add/drop multiplexing capability, and advanced bandwidth optimization and management.
A new architecture called trans-metro optical switching can address many of SONET's performance challenges. Using this technology, carriers have the tools needed for transitioning SONET to a data-centric traffic model.
Mass commoditization of basic services no longer works in the telecommunications business. Customers know that if one carrier cannot deliver on today's transport value propositions, deregulation provides them with the means to find another carrier that can. Businesses have learned the rules of this new game well and their service-level agreements (SLAs) with carriers prove it. To survive and remain competitive, carriers need a cost-effective and flexible infrastructure that leverages SONET's potential to deliver innovative services in this era of mass customization.
Experienced carriers know the massive transport scalability that can be achieved using SONET and the multitude of services it can deliver. Despite what certain industry pundits say about its capabilities, the root cause of transport problems and gridlock in today's networks is not due to the SONET protocol itself; the problems stem from the equipment used to process voice, data, and streaming media traffic.
Carriers operating MANs are especially sensitized to these equipment shortcomings. What carriers dream of is data-optimized, multiservice optical gear that will easily integrate into their existing network topologies and help to transcend network gridlock. At the same time, carriers want to wring the remaining voice-traffic revenue from TDM circuit switches and other mission-critical legacy equipment.
SONET currently operates at line rates ranging from DS-1 (1.5 Mbits/sec) to OC-192 (10 Gbits/sec), regardless of network size. OC-768 (40-Gbit/sec) transport systems are on the way. SONET can also transport information mapped into all prevailing logical-layer protocols. For voice, data, and video applications where transmission time is not critical, traffic can be transported within ATM cells or frame-relay packets that ride on top of the SONET optical signal. Even with these available options, TDM equipment that processes traffic in 64-kbit/sec (DS-0) slices is still used to handle approximately 80% of all traffic mapped onto SONET networks. However, this equipment was never designed to process and groom complex data types. The inherent limitations of circuit switching are compounded by those imposed by the telecommunications payload hierarchy and conventional SONET equipment.
First-generation SONET networks attempted to overcome some of these problems. These networks were built using add/drop multiplexers (ADMs), but this gear was expensive, inflexible, and took up considerable central-office rack space. The ADMs were designed to improve the efficiency of circuit switching; these devices supported the traditional telecommunications electrical-payload hierarchy up to DS-3 (44.736 Mbits/sec) and the optical interface hierarchy up to OC-48 (2.5 Gbits/sec).
Regardless of composition and requirement, traffic processed by these ADMs had to fit into a specific bandwidth slot whether or not it used the full bandwidth allocation. In cases where payload envelopes had to be extended to support dense data transfers, signal concatenation schemes were clumsy and granularity options coarse. Besides these limitations, this SONET equipment offered no support for local-area-network (LAN) protocols-a situation that is especially problematic considering that 10/100-Mbit/sec Ethernet is now ubiquitous and Gigabit Ethernet (1 Gbit/sec) is quickly gaining acceptance for enterprise backbone traffic (see Table). These systems were also Layer 1 devices. Therefore, the systems did not offer statistical multiplexing capabilities for traffic over-subscription, which requires equipment capable of Layer 2 and Layer 3 switching.
These limitations created problems in carriers' central offices for the incumbents and for the early competitive carriers who needed to colocate equipment in the incumbent's facility or in a carrier hotel. The equipment architectures, payload hierarchies, concatenation schemes, and the nonexistent support for LAN traffic all contributed to network gridlock.
To overcome first-generation SONET gear's limitations, new ADMs emerged. These systems shrunk physical unit size through use of application-specific integrated circuits (ASICs) instead of using discrete chips to perform different system functions. This gear was still designed for circuit-centric switching environments. Thus, it did not effectively support data traffic. In the mid-1990s, incumbent and aspiring carriers realized that achieving a data-centric transport model could not be accomplished by force-fitting SONET into the existing telecommunications infrastructure. Some carriers had the foresight and fortitude to encourage telecommunications chip and system vendors to develop new gear that would more fully optimize SONET's potential.
The outcome was "data-aware SONET" equipment that first ap peared in 1998. This gear offered statistical multiplexing that supported traffic over-subscription. Yet, these systems offered limited ADM functionality and their performance peaked at OC-12 (622-Mbit/sec) line rates that left them bandwidth-stunted in higher-traffic environments.
Despite the numerous advances in ADM, data-awareness, and statistical multiplexing, carriers and system developers knew that to meet demands in a data-centric world, a major evolutionary leap in SONET equipment's system architecture was needed.
Using the SONET protocol at the physical transport layer was non-negotiable. In the United States alone, hundreds of billions of dollars had already been spent on SONET in wide-area and metropolitan-area networks, and deployment continues. Solutions that would transcend network gridlock had to be deployable in the carrier's equipment rack and the gear had to perform at line rates of OC-192 to achieve bandwidth parity with the core networks.
The trans-metro market is where metro-access, metro-transport, and metro-core functions and services converge. Trans-metro optical-switching technology features a nonbinding and unconstrained architecture that distills the myriad functions of multiple SONET platforms into one. It offers scalability from DS-1 to OC-768, full-function ADM capability, and an advanced level of bandwidth optimization and management (see Figure 1).
In addition to performing digital-crossconnect (DCS) and ADM functions, trans-metro optical switches support all prevailing packet and circuit transport protocols in use today. Packet support encompasses ATM service access multiplexers/switches, frame-relay access switches, and Multiprotocol Label and IP switches (MPLS/IP). Circuit-switching support is provided for voice and private-line point-to-point circuits. Thus, customers' investments in LAN protocols (Ethernet, ATM, frame relay, IP) and TDM services are preserved.
The migration to a true trans-metro space follows several paths. Within existing MAN topologies, service-access multiplexers can be replaced with trans-metro optical switches at the ingress/egress points of OC-12 to OC-192 local loops and at OC-48 to OC-192 MAN edges.
Trans-metro optical switches support SONET line rates up to OC-768 and at the same time eliminate the discrete network elements previously required at each OC-n add/drop point. This switching technology is also used to interconnect the MAN edge with the network core. Here, core ATM switches groom aggregated traffic for transport over long-haul OC-48/192 dense wavelength-division multiplexing (DWDM) spans.
By deploying data-optimized, trans-metro optical-switching platforms at these interconnection points, carriers are able to address immediate needs and futureproof their networks. This way, technology investments are preserved, the MAN hierarchy becomes flatter, and trans-metro provisioning is dramatically simplified. Since trans-metro optical switches offer 4:1 statistical multiplexing, bandwidth is no longer wasted in point-to-point SONET links (see Figure 2).
Figure 2. By deploying data-optimized, trans-metro optical-switching platforms at network interconnection points, carriers' existing technology investments are preserved, the metro-area-network hierarchy is flattened, and provisioning is simplified.
With trans-metro optical switching, the entire SONET network hierarchy between the customer premises, central office, and major switch sites/ interexchange carriers' (IXCs') points-of-presence (PoPs) can be streamlined. This technology is also viable as premises equipment for enterprise customers seeking to deploy a true OC-n multiservice transport topology, due to its low cost, versatility, and provisioning elasticity.
Regardless of whether trans-metro optical switches are installed as customer-premises equipment, within ring or mesh topologies, or between the network edge and major switch/ IXCs' PoPs, connections can be established that support all prevailing cell and packet traffic, including ATM, frame relay, point-to-point protocol, IP, and LAN.
What carriers want is data-optimized, multiservice optical gear that integrates seamlessly into existing infrastructures and helps to futureproof past equipment investments. Trans-metro optical switching will allow carriers to extract the remaining value from aging TDM infrastructures, while transcending the gridlock that has impeded migration to data-centric networks.
Rafat Pirzada is vice president of marketing and founder of Cyras Systems Inc. (Fremont, CA). The company's Website is www.cyras.com.
Adopting dense wavelength-division multiplexing (DWDM) in metropolitan area networks (MANs) is not an immediate game-changing imperative.
Industry research indicates that between now and 2003, OC-48 (2.5-Gbit/sec) line rates will predominate as the workhorse, physical-layer, transport medium. Yet by 2005, OC-192 (10-Gbit/sec) and OC-768 (40-Gbit/sec) deployment will double. The turn-up of OC-192 in MAN rings will be explosive with OC-192 eclipsing lower line-rate transport systems as the baseline standard, according to Communications Industry Researchers (Charlottesville, VA).
Among long-haul backbone carriers, 1998 showed SONET deployment mixed among networks using OC-48 to OC-192 add/drop multiplexers (ADMs) with multiple lambda DWDM capabilities. In 1999, deployment shifted toward use of OC-192 ADMs with 16- to 40-lambda DWDM. Between long-haul and the metro areas, this mix of capacity will serve customer needs into the foreseeable future.
Most MAN carriers recognize that DWDM will, at some point, become an integral part of their network architectures. But based on existing pricing models within the metro market, the cost of deploying DWDM is still prohibitive, especially when rolling it out means retiring the existing network infrastructure. At the present time, such barriers make this an unacceptable option for metro carriers since approximately 80% of these companies' capital is allocated to optimizing central-office equipment, with the remainder used for channel-related outlays where DWDM makes more sense.
Since seamless end-to-end service provisioning is the goal for any network operator, deriving the most benefit from the existing SONET infrastructure within the metro transport market is a priority. So instead of deploying DWDM, these carriers are focusing on how to transition from a circuit-centric to a packet-centric transport model. Metro carriers are aggressively searching for creative SONET-based approaches that will efficiently and cost-effectively serve the local customer's circuit-switched requirements and at the same time build mixed-traffic on-ramps to the core networks where DWDM now predominates.