As DWDM technology migrates to metro networks, it will create a protocol- and bit-rate-independent optical layer that allows service providers to transparently offer broadband services.
Dense wavelength-division multiplexing (DWDM) technology has revolutionized backbone networks by enabling huge increases in the leveraged capacity of a single fiber. This, in turn, has significantly reduced gigabit-per-second per-kilometer transmission costs. Backbone capacity is more than doubling every year, easily outstripping Moore's Law. Fiber capacity is approaching terabits per second, and incremental bandwidth costs pennies per DS-0 (64 kbits/sec) per thousand kilometers. These developments have led networking gurus to declare the "death of distance."
A similar paradigm shift is expected in metropolitan networks. The value of DWDM implementation in this space, however, is fundamentally different. Metro service providers face network challenges beyond capacity, scalability, and transmission cost-per-bit. These carriers need to address service creation, rapid deployment, changing connectivity requirements, and cost containment in a highly competitive environment.
DWDM technology offers a solution that can meet these challenges. It can provide rapid capacity scaling, elastic bandwidth through bit-rate and protocol-independence, and the cost-effective delivery of new services.
Metro network engineers typically consider DWDM to address applications such as interoffice facilities, metropolitan collector rings, and business access. The metro network is optimized for the delivery of voice (time-division multiplexing--TDM) and private-line services such as DS-1 (1.544 Mbits/sec) and DS-3 (44.736 Mbits/sec).
The traditional methodology used to deliver multiple services has meant overlays or special assemblies to support each unique offering on a separate network (see Fig. 2a). With this infrastructure, the service provider has to contend with the following inefficiencies:
- Increasing cost for the overlay equipment
- Inefficient use of fiber
- High operational costs
- Inability to leverage economies of scale
- Management of multiple overlay networks.
Separate overlay networks result in high costs for carriers and long lead times for the turn-up of services.
Metro networks using DWDM technology can provide a flexible, scalable means to support a broad range of services for business access and interoffice applications. DWDM allows service providers to take advantage of the economies of scale associated with a single network that can support rapid turn-up of services. This technology enables carriers to offer a range of enterprise services and improve enterprise performance by extending local-area-network (LAN)-based services directly into the metro network, avoiding conversion to metropolitan-area-network (MAN)-based protocols (see Fig. 2b).
Carriers can also upgrade a service from a lower to a higher rate without altering the network infrastructure. The transparency of the DWDM network allows a wavelength delivering an OC-3 (155.52 Mbits/sec) today to migrate to OC-12 (2.5 Gbits/sec) or OC-48 (10 Gbits/sec) "on-the-fly" without the need for a field visit or additional circuit packs. The cost savings, revenue enhancement, and fast provisioning will provide a competitive advantage to carriers that successfully exploit DWDM technology.
For DWDM networks to support current and emerging metro applications, these infrastructures must deliver on a broad range of service interfaces. Escon, a native 200-Mbit/sec protocol, is used to connect mainframes to other mainframes or to storage devices. Initially, this protocol was confined to device interconnection in the computer room. With the need to protect services for business continuity, however, there is a requirement to extend this protocol across geographically dispersed locations.
Currently, to extend tariffed Escon services for remote connectivity across the MAN, Escon must be throttled down to a DS-3 connection. To accomplish this protocol conversion, the data-assurance protocols inherent in Escon must be discarded, leaving errors undetected. With business customers' growing reliance on mission-critical data and the move to 24-hour e-commerce strategies, this alternative is no longer viable. Information-technology (IT) mangers seek solutions from service providers that allow Escon to be carried in its native format to ensure full-protocol capability and data integrity.
In the future, Escon will migrate to Ficon, a higher-speed derivative, which operates at 1 Gbit/sec for Escon applications. The DWDM network enables service providers to meet the Escon service demand at wire speed today with the capability to upgrade to Ficon when the need arises, without replacing expensive transport equipment.
Intersystem coupling at 1.05 Gbits/sec and external timing reference at 16 Mbits/sec are used together to enable parallel processing or cluster computing. The application for these protocols is mainframe redundancy so that the failure of one computer will not result in downtime. Once again, the deployment of these services across a protocol and bit-rate-independent DWDM network allows the service provider to cost-effectively provide these services across the MAN.
Another data service, Fibre Channel, is used by a host of IT storage devices to backup and restore applications. The need to network these protocols across the MAN is similar to the requirements for Escon/Ficon extensions--to enable business continuity if a catastrophic event occurs at the primary data center.
Once relegated to the LAN environment, Fast Ethernet and Gigabit Ethernet can now reach across the MAN in native form with the DWDM network. In this configuration, these protocols offer an efficient, reliable means of interconnecting IP devices. The protocols do not require that data be mapped into fixed bit-rate-intermediary formats before it is transported. The DWDM network's bit-rate and protocol independence allows the service provider to extend native data-rate Ethernet LANs into MAN applications, while maintaining the capability to migrate from Fast Ethernet to Gigabit Ethernet capabilities through a simple point-and-click network-management interface.
SONET/SDH is the standard technology used to provide communication services across metro and regional networks because of its ability to multiplex low-rate services (DS-1/3, E1) and because of the usefulness of the standardization of interconnect protocols across multiple vendors. For these reasons, SONET/SDH will continue to be deployed for traditional services. The DWDM network, however, will enable service providers to efficiently upgrade these services to higher bit rates (i.e., OC-3 to OC-12 to OC-48) and to simultaneously accommodate native data-rate services over the same infrastructure.
Customer applications for video transmission increasingly require extremely high bandwidth to transport uncompressed, studio-quality video at up to 270 Mbits/sec. Rather than isolating this high-value, time-sensitive traffic on a separate overlay network, service providers need to reduce costs and management complexity by transmitting D1 video at its native date rate. With a DWDM network, service providers can transmit D1 video and any other protocol at wire speed though a common universal network, thereby attaining economies of scale. In the future, when high-definition television (HDTV) is deployed at 1.48 Gbits/sec, the DWDM network can support it with point-and-click provisioning.
Figure 3 is a simplistic diagram of today's metro transmission network. A DS-1/3 or OC-N circuit routes signals through edge concentration elements (SONET/SDH rings), which are then handed to a large SONET/SDH crossconnect. The crossconnect performs connection management and grooming for the access and interoffice transmission devices located in the office. It also provides connectivity to local network-processing gear such as routers and ATM switches.
The physical architecture of the optical domain is similar in many respects to today's transmission networks, due to the common physical plant and service demand. The large number of interoffice sites that require connectivity and survivability necessitate an optical ring; most of the SONET/SDH deployed in this portion of the network today is ring-based. The access portion of the network requires both a low-end ring and a point-to-point system. Connectivity and grooming requirements for optically based services will necessitate optical crossconnects for larger offices once critical mass is reached.
Open management interfaces are vital to the deployment and utilization of all networks, and DWDM is no exception. Integrated management of the multiple bit-rate- and protocol-independent DWDM environment is accomplished through 3R regeneration, performance monitoring (PM), and digital wrappers for fault isolation.
Although protocol and bit-rate transparency is a key advantage of metropolitan DWDM networks, photonic transparency can actually make these infrastructures difficult to engineer and turn up. Engineering a photonically transparent connection requires attention to numerous fiber characteristics such as attenuation, dispersion, noise, nonlinear distortions, and jitter. In today's SONET/SDH systems, these factors are addressed at the line layer; thus, individual SONET/SDH line systems are engineered independently and combined in a network. Similarly, long-haul DWDM systems are point-to-point topologies; all wavelengths have the same endpoints and can be engineered as a group.
By definition, each connection in a metro DWDM network can have different endpoints. If every connection is photonically transparent, then each one must be engineered separately. This approach defeats one of the key benefits of DWDM networks, which is the ability to rapidly introduce services, including connections with new endpoints. For example, if a transparent connection from point A to point B is rerouted to point A to point C, the system must be re-engineered. The new connection may require the deployment of an optical amplifier, which in turn will force the re-engineering of other photonically transparent connections also affected by the amplifier.
This problem is resolved by deploying photonically transparent subnetworks, the boundaries of which are defined by protocol- and bit-rate-independent 3R regeneration. 3R regeneration encompasses the following functions:
- Regeneration ensures that the outgoing power level of the connection is adequate to cover the next hop
- Reshaping removes pulse-shape distortions such as those caused by dispersion
- Retiming removes time-domain distortions of the digital pulse so that downstream clock-recovery circuits can properly receive the signal.
Initially, these photonically transparent subnetworks will be the ring and point-to-point transport systems that interconnect subtending devices. As more of these systems are deployed, optical-crossconnect systems will be used to provide flexible and manageable points of interconnection. These systems must have the ability to provide 3R regeneration for each connection to avoid creating a network where the costs of planning, engineering, and provisioning become prohibitive.
At the customer-demarcation point in the network, PM must be invoked to arbitrate service-level agreements. This approach requires an optical layer that is protocol aware and able to read protocol PM data. Thus, the optical layer must have the ability to read, store, and interpret multiprotocol PM information at network interfaces. If SONET were presented, for example, the optical-layer interface must read the SONET "B1" and "J0" bytes to obtain path performance and trace information at the DWDM network endpoints. Similarly, if Gigabit Ethernet passes across the optical layer, the B4/B5 multiplex code requires error monitoring.
Protocol- and bit-rate-independent DWDM does not eliminate the need to do fault isolation or guarantee performance at hand-off points between service providers and customers. Fault isolation and PM techniques have evolved based on digital primitives that support the detection of individual bit errors and the derivation of a variety of alarms, alerts, and performance parameters. Although sophisticated monitoring techniques are available for the photonic domain, these methods do not produce the same granularity and accuracy as digital approaches.
Early long-haul DWDM systems were deployed to relieve fiber constraints for stacked SONET/SDH systems. Since the SONET/SDH and DWDM terminals tend to be colocated, the SONET/SDH terminals provide the end-to-end performance monitoring for the DWDM systems using the primitives built into the SONET/SDH signal. Intermediate regeneration points perform nonintrusive monitoring of the SONET/SDH signal to provide fault isolation.
One of the key benefits of DWDM networks, particularly in metropolitan applications, is the ability to transport client signals other than SONET/SDH. Many, if not all of these signals lack physical-layer PM primitives. Therefore, the long-haul approach does not work. It is also cost-prohibitive to manage the optical layer by processing lower-layer protocols. Even if the signals did have PM primitives, it is impractical for the DWDM system to nonintrusively monitor every protocol, especially when it is not known what protocols the system will carry.
Therefore, a technique known as the digital wrapper is required. A digital wrapper is essentially a frame that encapsulates the client signal and adds overhead (integrity) information. DWDM equipment can use the digital wrapper for fault isolation and PM regardless of the client signal's protocol.
In order to provide end-to-end PM for the client signal, the digital wrapper must accompany it across the entire DWDM network. Therefore, the digital wrapper is an integral part of the interface that interconnects DWDM subnetworks. Since these subnetworks may reside in different service-provider networks with equipment supplied by different vendors, the subnetwork requires an open interface defined by industry standards.
Early deployments of optical-transport-networking (OTN) equipment will transport SONET/SDH client signals and enterprise network applications where the protocol is not supported by SONET/SDH (e.g., Gigabit Ethernet, Fibre Channel, Escon).
The enterprise applications tend to involve a single service provider and are usually confined--by the application--to a metro region, often within a single OTN subnetwork, which reduces the need, in the short term, for sophisticated PM capabilities.
The SONET/SDH applications' PM requirements are best addressed, in the short term, by the use of nonintrusive monitoring of the SONET/SDH primitives by the OTN equipment, similar to the approach currently used in long-haul networks. This type of monitoring will result in performance parameters that the industry is accustomed to and that can be easily communicated to legacy-network surveillance systems.
Pre-standard, proprietary digital-wrapper techniques may result in performance parameters that are difficult to integrate into today's surveillance systems. Moreover, this integration will have to be repeated once the standard version is available. Even more important, a proprietary version does not address one of the key requirements of the digital wrapper--end-to-end PM over multivendor networks.
The protocol and bit-rate transparency of the deployed OTN equipment will facilitate upgrades to standards-compliant, digital-wrapper-based PM when standards are established (see Fig. 5).
The International Telecommunication Union-Telecommunication standards body is in the process of defining international digital-wrapper standards with strong input from the United States via the American National Standards Institute-accredited Technical Subcommittee, T1X1. Current activities include the format, overhead allocations, and client signal-adaptation mechanisms. With the current pace of activities, standards-compliant equipment is likely to emerge later this year. As with any networking technology, creation of and adherence to standards are key.
DWDM networks will enable service providers to meet-and drive-the changes happening in the metro. The scalability and flexibility inherent in bit-rate- and protocol-independent solutions can deliver new services and better network economics. To realize the full potential of this technology, metro DWDM networks must be easy to manage, engineer, and provision, therefore requiring standards-based implementation of 3R regeneration, PM, and digital wrappers.
With these capabilities in place, the changes seen in long-haul networks will be reflected in the metro. The "death of distance" will be coupled with the "birth of broadband services" ubiquitously across the metro.
Brent Allen is the director, brand management, for OPTera Solutions and James Rouse is the director of market development for the OPTera Solutions division of Nortel Networks (Ottawa, Ontario, Canada).