SAP details: Events News - Training - World wide sites  - Sap Labs SAP Magazine  - SAP Screen saver - Stock Price
WIDE AREA NETWORK    (article by Kaushik Das)

High Speed Networking & Communications with WAN Technologies

Introduction: With the rapid growth of IP and the Internet, enterprise WAN networking is undergoing rapid change. New service offerings, such as Frame Relay and Asynchronous Transfer Mode (ATM), provide new options to complement existing leased-line private networks. The increasing volume of data traffic, increasingly based on IP, has shifted the WAN from traditional time-division multiplexers to a routed architecture. Concurrently, voice over IP, voice over Frame Relay, and voice over ATM enable data/voice integration on these new services. However, recurring service cost is still the dominant issue in enterprise WAN budgets, with labor costs to operate and manage the WAN also a significant issue.

As the leader in routed WAN and IP-based networking for enterprise networks, Cisco brings new capabilities to enterprise routed WAN networks to support new services, reduce costs, and enable enterprise WAN networks to grow with the demands of enterprise network applications. This white paper briefly discusses the evolution of enterprise WAN networking and describes how enterprise network managers can utilize the new multi-channel networking capabilities, advanced routed software services, and new features on the Cisco 7500 and 7200 routers.

At the most basic level, a wide-area network exists to interconnect individuals or communities in a useful and cost-effective manner. Service providers deploy hardware, software, and people to maximize the utility of their network while minimizing its capital and operational cost. The utility of the network comes from the services it enables: how well each service solves a customer problem, how quickly a service can be deployed to a customer, and how reliably the service performs. Of course, the utility of the deployed services determines the revenue realized by the WAN. The cost of a network comes from the infrastructure deployed to support it: the capital cost of required equipment, the cost of supporting information systems, the cost of support personnel, the network bandwidth consumed by the services, and so on. Obviously, the most profitable WAN service providers are those who are best able to deploy their revenue generating services at the lowest costs.

To help optimize revenue-generation capabilities and network cost, service provider networks often include two layers: edge and core. The "Edge" of the network is used to deploy revenue-generating services. The "Core" is used to optimize cost by providing efficient transport and bandwidth optimization of edge-provided traffic. Although it is an oversimplification to say "Edge = Revenue Maximization" and "Core = Cost Minimization"---after all, service providers spend the bulk of hardware budgets on edge platforms---it is useful to understand the important distinctions between edge and core functions. This white paper addresses the roles played by both edge and core platforms in wide-area networks, looks at key drivers of change that affect both, and describes Cisco's vision of next-generation edge and core requirements. Finally, it discusses Cisco edge and core platforms, providing pointers for product-specific detail.
Introduction to WAN Technologies

This topic introduces the various protocols and technologies used in wide- area network (WAN) environments. Topics summarized here include point-to-point links, circuit switching, packet switching, virtual circuits, dialup services, and WAN devices.

Figure 3-1: WAN technologies operate at the lowest levels of the OSI model

What is WAN?                       

A WAN is a data communications network that covers a relatively broad geographic area and often uses transmission facilities provided by common carriers, such as telephone companies. WAN technologies function at the lower three layers of the OSI reference model: the physical layer, the data link layer, and the network layer.

Point-to-Point Links

A point-to-point link provides a single, pre-established WAN communications path from the customer premises through a carrier network, such as a telephone company, to a remote network. A point-to-point link is also known as a leased line because its established path is permanent and fixed for each remote network reached through the carrier facilities. The Carrier Company reserves point-to-point links for the private use of the customer. These links accommodate two types of transmissions: datagram transmissions, which are composed of individually addressed frames, and data-stream transmissions, which are composed of a stream of data for which address checking occurs only once. Figure 3-2 illustrates a typical point-to-point link through a WAN.

 Figure 3-2: A typical point-to-point link operates through a WAN to a remote network

Circuit Switching 

Circuit switching is a WAN switching method in which a dedicated physical circuit is established, maintained, and terminated through a carrier network for each communication session. Circuit switching accommodates two types of transmissions: datagram transmissions and data-stream transmissions. Used extensively in telephone company networks, circuit switching operates much like a normal telephone call. Integrated Services Digital Network (ISDN) is an example of a circuit-switched WAN technology, and is illustrated in Figure 3-3.

Figure 3-3: A circuit- switched WAN undergoes a process similar to that used for a telephone call.

Packet Switching

Packet switching is a WAN switching method in which network devices share a single point-to-point link to transport packets from a source to a destination across a carrier network. Statistical multiplexing is used to enable devices to share these circuits. Asynchronous Transfer Mode (ATM), Frame Relay, Switched Multimegabit Data Service (SMDS), and X.25 are examples of packet-switched WAN technologies (see Figure 3-4).

Figure 3-4: Packet switching transfers packets across a carrier network.

WAN Virtual Circuits  

A virtual circuit is a logical circuit created to ensure reliable communication between two network devices. Two types of virtual circuits exist: switched virtual circuits (SVCs) and permanent virtual circuits (PVCs). SVCs are virtual circuits that are dynamically established on demand and terminated when transmission is complete. Communication over an SVC consists of three phases: circuit establishment, data transfer, and circuit termination. The establishment phase involves creating the virtual circuit between the source and destination devices. Data transfer involves transmitting data between the devices over the virtual circuit, and the circuit-termination phase involves tearing down the virtual circuit between the source and destination devices. SVCs are used in situations, in which data transmission between devices is sporadic, largely because SVCs increase bandwidth used due to the circuit establishment and termination phases, but decrease the cost associated with constant virtual circuit availability. A PVC is a permanently established virtual circuit that consists of one mode: data transfer. PVCs are used in situations in which data transfer between devices is constant. PVCs decrease the bandwidth use associated with the establishment and termination of virtual circuits, but increase costs due to constant virtual circuit availability.

WAN Dialup Services
Dialup services offer cost-effective methods for connectivity across WANs. Two popular dialup implementations are dial-on-demand routing (DDR) and dial backup. DDR is a technique whereby a router can dynamically initiate and close a circuit-switched session as transmitting end station demand. A router is configured to consider certain traffic interesting (such as traffic from a particular protocol) and other traffic uninteresting. When the router receives interesting traffic destined for a remote network, a circuit is established and the traffic is transmitted normally. If the router receives uninteresting traffic and a circuit is already established, that traffic also is transmitted normally. The router maintains an idle timer that is reset only when interesting traffic is received. If the router receives no interesting traffic before the idle timer expires, however, the circuit is terminated. Likewise, if uninteresting traffic is received and no circuit exists, the router drops the traffic. Upon receiving interesting traffic, the router initiates a new circuit. DDR can be used to replace point-to-point links and switched multi-access WAN services.

Dial backup is a service that activates a backup serial line under certain conditions. The secondary serial line can act as a backup link that is used when the primary link fails or as a source of additional bandwidth when the load on the primary link reaches a certain threshold. Dial backup provides protection against WAN performance degradation and downtime.

WAN Devices

WANs use numerous types of devices that are specific to WAN environments. WAN switches, access servers, modems, CSU/DSUs, and ISDN terminal adapters are discussed in the following sections. Other devices found in WAN environments that are exclusive to WAN implementations include routers, ATM switches, and multiplexers. 

WAN Switch

A WAN switch is a multiport internetworking device used in carrier networks. These devices typically switch such traffic as Frame Relay, X.25, and SMDS and operate at the data link layer of the OSI reference model. Figure 3-5 illustrates two routers at remote ends of a WAN that are connected by WAN switches.

Figure 3-5: Two routers at remote ends of a WAN can be connected by WAN switches.

 Access Server 

An access server acts as a concentration point for dial-in and dial-out connections. Figure 3-6 illustrates an access server concentrating dial-out connections into a WAN.

Figure 3-6: An access server concentrates dial-out connections into a WAN.


A modem is a device that interprets digital and analog signals, enabling data to be transmitted over voice-grade telephone lines. At the source, digital signals are converted to a form suitable for transmission over analog communication facilities. At the destination, these analog signals are returned to their digital form. Figure 3-7 illustrates a simple modem-to-modem connection through a WAN.

Figure 3-7: A modem connection through a WAN handles analog and digital signals.  


A channel service unit/digital service unit (CSU/DSU) is a digital-interface device (or sometimes two separate digital devices) that adapts the physical interface on a data terminal equipment (DTE) device (such as a terminal) to the interface of a data circuit-terminating (DCE) device (such as a switch) in a switched-carrier network. The CSU/DSU also provides signal timing for communication between these devices. Figure 3-8 illustrates the placement of the CSU/DSU in a WAN implementation.

Figure 3-8: The CSU/DSU stands between the switch and the terminal.

 ISDN Terminal Adapter

An ISDN terminal adapter is a device used to connect ISDN Basic Rate Interface (BRI) connections to other interfaces.

Figure 3-9: The terminal adapter connects the ISDN terminal adapter to other interfaces.

Such as EIA/TIA-232. A terminal adapter is essentially an ISDN modem. Figure 3-9 illustrates the placement of the terminal adapter in an ISDN environment.

In-Depth of WAN Technology:

The WAN services market is large and is growing rapidly. With the rapid growth of the Internet and IP/intranet applications in enterprise LANs, IP-based WAN services are growing at a Compound Annual Growth Rate of 73 percent. This scenario is placing an increasing spotlight on the routed WAN infrastructure in enterprise WAN’ s and a focus on IP-based networking services and software features. In addition, the performance increases in the LAN are driving up the speeds (and costs) of WAN connections for enterprises. Figure 1 illustrates this rapid growth in the WAN services market.

Figure 1: Growth in US WAN Services Market

It is expected that leased-line services will continue to make up a very large part of the market, with Frame Relay also growing at a rapid rate. (See Figure 1.) Many different service offerings will continue to coexist, as each has advantages in different applications with different cost versus control tradeoffs.

Terms related with WAN:

Leased Lines

Leased lines are the largest WAN service category today. These services provide dedicated, private bandwidth with low and fixed delay characteristics, with speeds from 56 KBPS through 155 MBPS Synchronous Optical Network/Synchronous Digital Hierarchy (SONET/SDH) fiber. Leased-line services provide a good choice for networks that require a proven service with a high degree of control over the network, a common requirement in large enterprise WAN networks. This service type is also commonly used as the "last-mile" technology to access other services (for example, Internet or Frame Relay). It is the most expensive WAN service alternative in dollars per bits per second, but it also provides the greatest network control.

Frame Relay

Frame Relay provides virtual circuit connectivity for enterprise networks that require 56 KBPS up to T1/E1 speeds. It costs less than leased lines because it uses statistical multiplexing of packets to gain efficiencies within the network, at the cost of a less-stringent bandwidth and latency guarantee. Frame Relay is being widely deployed in enterprise networks to connect regional and branch offices into the enterprise backbone.


Like Frame Relay, ATM provides virtual circuit connectivity but at higher speeds (DS3/E3 up to OC-3/Synchronous Transport Module level 1 (STM-1) today, with support for T1/E1 in the future). It is capable of higher quality of service (QoS) than Frame Relay and is used today to connect enterprise WAN and MAN backbone sites.


This networking technology is still applicable in low-speed environments where circuit noise is a problem. It is typically used in international networks in places where the network infrastructure cannot yet support Frame Relay or ATM.

IP Access

IP Access is the fastest growing WAN service and today is primarily used for connections to the public Internet. Increasingly, however, enterprise network managers are looking to IP access as a way to outsource some of the management of the enterprise WAN Intranet. New IP protocol technologies for security, QoS, and virtual private networking will enable IP access to become a viable option for leveraging public data networks for private enterprise WANs.

Enterprise networks have traditionally combined multiple applications on time-division multiplexing (TDM)-based leased-line WAN backbones. For the same cost-saving reasons, they want to achieve this same integration across newer packet- and cell-based WAN services. The key to supporting this data/voice/video integration is support for QoS in the service offering and in the enterprise WAN router that connects to the WAN service. This has been a characteristic of Cisco routers for T1/E1 speeds (for example, support for priority queuing and custom queuing), but is increasingly needed at higher speeds.

Finally, it is worth noting that recurring bandwidth costs still dominate the budgets of enterprise WAN networks. Leased lines are still a dominant technology for WAN networking today, providing control and guaranteed bandwidth to enterprise customers. As enterprise WAN networks grow, more cost-effective leased-line termination solutions are required for the enterprise WAN router. In addition, increasingly sophisticated software and services in WAN routers that can manage bandwidth and measure, analyze, and account for traffic is necessary to improve the operational efficiency of the WAN, thereby lowering costs.

These trends lead to the following requirements for enterprise WAN equipment:

·         Multiple WAN services: WAN devices need to provide support for a broad range of WAN services, enabling the enterprise to pick the best price and capabilities match to their requirements, whether it is leased-lines, IP access, Frame Relay, ATM, X.25, or a combination of services.

·         Speeds greater than T1/E1: WAN equipment needs to scale with the increasing performance demands of enterprise networks and the capabilities of WAN services. This scenario is obviously important for ATM networks, supporting up to OC-3/STM-1 speeds, but is important for leased-line backbones as well.

·         Bandwidth Control: In order to combine multiple applications with differing requirements on the same WAN service, the service and the enterprise WAN equipment must provide integrated and compatible priority and QoS capabilities. This QoS capability must also support high-speed WANs as well as a range of WAN services to meet the increasing performance requirements of enterprise Intranets.

·         Reduce equipment cost: In many enterprise WAN solutions today, several pieces of equipment are required to deliver the solution. WAN equipment must integrate more functions into a single platform and provide higher-density WAN interfaces to save costs.

·         Reduce management cost: WAN equipment must reduce the complexity of building enterprise WAN networks and provide improved management visibility into the network and its operation.

Today, the Cisco 7500 router is the strategic high-end router for both collapsed backbone LAN and enterprise WAN applications in most of the world's enterprise networks. Over the next few years, multilayer switching technologies and products, such as those developed by Cisco for the Catalyst family of LAN switches, will provide increased Layer 3 performance in campus LANs to address the needs of new Fast Ethernet and Gigabit Ethernet infrastructures. The Cisco 7500 provides the evolving Enterprise LAN, with its shift to fewer media types, with seamless integration of existing multiprotocol and multiple media networks such as Fiber Distributed Data Interface (FDDI), Token Ring, and ATM. It also provides integration into the data center with channel-attached Enterprise System Connection (ESCON) and bus and tag connections to the mainframe. In addition, the Cisco 7500 remains an outstanding LAN solution for those networks that do not yet need the higher performance of multilayer switches, or that have a conservative posture regarding the adoption of new technologies.

As this LAN shift occurs, the Cisco 7500 and 7200 routers will retain their strategic WAN role in enterprise networks. To meet the new requirements of service flexibility, high-speed connectivity, bandwidth management, and cost reduction in the enterprise route WAN, Cisco is significantly enhancing the capabilities of the Cisco 7500 and 7200 routers. The next sections describe how these new enhancements meet the emerging requirements of the new enterprise routed WAN network. 

ATM Connectivity

Where ATM WAN connectivity is the right choice, the Cisco 7500 and 7200 routers now support a family of WAN-capable ATM port adapters. With a choice of DS3, E3, OC-3/STM-1 single-mode (intermediate or long reach) or multimode interfaces, the enhanced ATM port adapter provides key ATM traffic shaping features needed to efficiently use ATM WAN services. This traffic shaping capability will also enable the new enhanced ATM port adapters to support ATM QoS across an ATM backbone, and Tag Switching QoS across a Tag Switching ATM backbone. Furthermore, the enhanced ATM port adapter hardware is available bit rate (ABR)-ready, and will support this new ATM technology in a future Cisco IOS software release. This scenario will enable network managers to take advantage of cost-effective high-throughput ABR services as they are offered by service providers.

Advanced Services

The third key to meeting the requirements of new enterprise routed WANs is high-speed support for advanced services. Supporting capabilities such as QoS and multicast, and providing the ability to measure and report on the network, will enable enterprise network managers to combine multiple applications onto the enterprise WAN to make efficient use of new WAN services and to manage the expense of scarce WAN resources.

High-Speed Quality of Service

Cisco IOS software today supports a wealth of QoS capabilities for differentiated network services such as priority queuing and custom queuing. These services are widely deployed in enterprise networks today, particularly in support of mission-critical applications. Historically, they have supported line speeds of up to T1/E1 rates and were not applicable at higher speeds. In 1997, Cisco announced Internet QoS features for Cisco routers. These innovative features enabled Internet service providers to offer IP QoS services to their customers, with support in their backbones at high speed.

These high-speed IP QoS capabilities are now available for enterprise WAN networks that utilize DS3, E3, or
OC-3/STM-1 connections. By classifying traffic as to its importance, the enterprise routed WAN can prioritize packets properly on high-speed WAN links when congestion occurs. This setup ensures that mission-critical application packets, or time-sensitive real-time traffic, are sped to their destinations. The new enterprise QoS capabilities include:

Committed Access Rate (CAR)

This feature performs both packet classification and bandwidth management functionality. The packet classification features let users partition network traffic into multiple priority levels or classes of service (CoSs). The network manager can define up to six CoSs using the three precedence bits in the type-of-service (ToS) field in the standard IP packet header. The manager can then use other QoS features to assign appropriate traffic-handling policies, including congestion management, bandwidth allocation, and delay bounds for each traffic class.

Weighted Random Early Detection (WRED)

This feature provides network managers with powerful congestion-control capabilities designed to provide preferential treatment for premium-class traffic under congestion situations while concurrently maximizing network throughput and capacity utilization and minimizing packet loss and delay.

Weighted Fair Queuing (WFQ)    

This feature provides bandwidth allocations and delay bounds to specified IP traffic sources by segregating the traffic into flows or classes and then servicing the various queues according to their assigned weights. WFQ classes can be defined by IP precedence, application ports, or incoming interface.

But fully utilizing QoS in a network requires much more than just queuing capabilities in the network nodes. CiscoAssure Policy Networking enables business users and applications to utilize the intelligence that is embedded in a network. Simply put, CiscoAssure Policy Networking makes it easier for a network manager to take advantage of distributed network intelligence features.

To set up a QoS policy, the network manager uses the CiscoAssure Policy Administration graphical user interface (GUI) to specify a policy based on business rules. A QoS policy binding is then created and activated by QoS policy servers and network-based enforcement in Cisco IOS devices. The Common Open Policy Service (COPS) Protocol provides policy exchange between the policy servers and the Cisco IOS software embedded in the intelligent network elements. Cisco IOS software translates the policy binding into local QoS enforcement mechanisms such WFQ or WRED. 

WAN Evolution: Edge and Core

 Service provider edge networks are responsible for delivering services. The word "service" is appropriate---the edge exists to serve the customer. Key functions of the edge network include creating and delivering services on a variety of physical interfaces, and providing service quality and differentiation. Service providers deploy core networks to serve their edge networks. Key functions of the core include high-capacity, high-speed, highly available transport of traffic. Because of a host of business- and technology-driven forces that are described in this paper, the roles of WAN edge and core networks must evolve. Edge networks must integrate IP functionality and support higher-speed interfaces for direct connection to dark fiber or Dense Wave Division Multiplexing (DWDM). Core networks must take advantage of rapidly advancing optical technology and become operationally transparent to edge-delivered services.

 Demand for Services and Bandwidth

Enterprise customers are continually demanding new types of services to solve specific problems; service providers are expanding their service portfolios to differentiate their offerings. In the early '90s, there was one type of WAN service---now called a private line. Soon the WAN service portfolio expanded to include Frame Relay service and private-line service. More recently, the WAN portfolio includes more specialized "flavors" of Frame Relay (such as SNA Migration) as well as connectionless services---typically Internet access. This trend of service proliferation is expected to continue for the foreseeable future, as service providers add IP virtual private networks, voice over IP-enabled Computer Telephony Integration, content hosting, and other services. Service proliferation also drives the need for more bandwidth as more users connect to public networks, and as individual applications become more bandwidth intensive.

Supply of Bandwidth

Given the ever-increasing demand for WAN bandwidth, it's certainly fortuitous that recent developments in DWDM technology have lowered the marginal cost of network bandwidth at OC-48/STM-16 and above. DWDM systems, which allow multiple wavelengths of light (each supporting a high-speed transmission link) to be carried on a single fiber, have shaken up the economics of long-haul bandwidth. In effect, carriers that deploy DWDM increase their fiber capacity 16-80 times, depending on the system deployed, at a cost that is (for most long routes) lower than the cost of deploying new fiber plant.

New Competitive Environment

The third driver of change in network edges and cores involves the new competitive environment. Worldwide telecommunications deregulation has been widely discussed, so that subject isn't examined thoroughly in this document, nor does this document attempt to retell the (related) story of the new breed of well-funded, data-oriented new entrants to the service provider industry. Instead, it focuses on three consequences of deregulation and new entrants, which affect the roles of edge and core switches in wide area networks. These include:

·         Emergence of higher value-add services---As new players enter service provider markets with competitive offerings, carriers are forced to drop prices on their existing WAN services, lowering margins. Amplifying this problem is the gradual commodification of WAN transport services (such as Frame Relay) as service providers attempt to match the competition feature-for-feature. To boost margins in this environment, service providers are looking to offer higher-value services (such as IP VPNs). Because most high-growth enterprise applications are IP based, the bulk of the higher-value services are IP-driven. In fact, in the first decade of the 21st century, IP-based WAN services will account for most of service provider WAN profits, even though IP-based WAN services will not yet account for the majority of service provider traffic.

·         Requirements for faster time-to-market for new services and faster time-to-turn-up for individual customers--- Increased competition is forcing service providers to differentiate along the dimension of speed. Not only is it important for some service providers to be "First with (Service X)", it's also important to offer service activation in less time than the competition. Enterprise WAN service deals can be won by the provider who can "turn up" service the fastest.

·         New economics of providing services driven by the lower cost basis of new service provider networks---The fundamental business premise behind most new service provider entrants has been "same or better service with lower infrastructural cost." To compete in this lower cost-basis environment, service providers need to get the most leverage out of their infrastructures---in other words, to generate the most revenue out of their deployed plant, support systems, and personnel. 

Betting on WAN Access Technology

Armed with Web browsers, SVGA monitors and 32-bit sound cards, corporate technology users send and receive information at rates never anticipated. Just a few years ago, three or four workstations would not have used the kind of bandwidth a single workstation devours today. Administrators feel the heat not only in their LANs, but especially in their WANs, many of which have been in place long enough to feel the brunt of the double-barreled
multimedia-Internet explosion. Network managers Stumble over outmoded long-term contracts with service providers and, worse, the inadequacies of their WAN technology. And as stacked, as the deck may seem against yesterday's technology sure bets, gambling on WAN technology that will carry corporations and users into the 21st century is even riskier.

One of the high-stakes arenas in the WAN technology crapshoot is last-mile access. Technologies such as analog, 56 KBPS, ISDN and even T1--the traditional high rollers in the game of "last mile" connectivity for branch/remote office communication--are reaching their cost or performance limits and are ripe for replacement by brash new players and technologies like the cable companies

with Hybrid Fiber-Coaxial (HFC) networks and the telephone companies with Asymmetric Digital Subscriber Line (ADSL) networks. These new technologies--which boast deeper bandwidth pockets and greater reliability at a better price--have been carefully planned, tested and fermented over the past decade.

Roll The Dice
What will these technologies mean for you and your business? Nothing--until we start seeing deployment over the next one to three years. Then these services will have at least limited availability except in the largest and most lucrative metropolitan areas.

Despite the proclamations from the cable industry, cable isn't going to be a business solution for some time because of its history and network structure. Enterprise networks need a broad portfolio of services, such as dial-up, private networks, frame relay and management. At the same time, many customers want simplified billing, consolidated contracts, service-level guarantees and performance monitoring. The cable companies cannot provide any of this. They do, however, have the potential to win the war for residential access beating out analog, ISDN and ADSL services, while driving exchange carriers to compete with one another and with the cable industry for customers.

ADSL will prevail as a temporary solution for business access between small-office/home-office (SOHO) sites and the corporate office. Telephone companies understand the need for Internet access and telecommuting and can use what they've learned from ISDN to help them prepare for the corporate rush. However, it won't be easy for the telcos either: Most of them still are stumbling over ISDN deployment and billing and how to handle the load 128-Kbps ISDN brings on their equipment.

The competition between these industries will drive down the price of both HFC- and ADSL-based networking. From the high end, it will also drive down the price of T1 access, possibly making it cheap enough to consider for branch-office and small-office connectivity. When the dust settles and services are installed, applications that demand high bandwidth, voice over IP and other services will be a reality at corporate sites and SOHO locations.

Hitting The ADSL Jackpot
To compete against local cable companies, competitive access providers and, in many instances, themselves, the local telephone companies have been testing ADSL technology, which is based on their existing copper infrastructure. Although its topology differs, ADSL resembles the cable companies' HFC network in that it uses frequency demodulation to carry services, such as video and data, to plain old telephone service (POTS) customers. Like cable, the service is dedicated and non-switched, and it requires a modem to demodulate the signals and tie the data and other services to the appropriate equipment at the customer site.

Telephone companies have an advantage over cable companies in that they do not have to upgrade switches or replace wiring to make ADSL work. All they need to do is place back-end modems at the local central office (CO) and remote-end modems at each customer's premises. Unlike ISDN (ADSL's sibling), no additional provisioning, equipment configuration or copper lines are necessary; like cable, implementation is simple, requiring little or no technical support to set up and operate.

Depending on the type of line-coding used to implement the ADSL network and the distance between the customer site and the CO, data transfers can reach speeds of 640 MBPS upstream and up to 9 MBPS downstream. The best line-coding method is a topic of debate in the ADSL industry. The most popular methods are Discrete Multitone (DMT) and Carrierless Amplitude Phase (CAP) modulation. Although IEEE chose DMT as the standard, many vendors continue to develop for CAP and other line-coding techniques, such as 2B1Q. Until recently, the idea was for carriers to choose a technology and for customers to provide compatible equipment. Now, the trend is shifting toward giving customers a choice of technology, with which the carriers must comply. As a result, customers will find that understanding the differences between line-coding techniques will become increasingly important (line speed and error correction depend on them).

ADSL's topology is similar to that of a star network, which usually doesn't share available bandwidth among everyone at the local CO (see "ADSL Network Access," on page 50). This makes an ADSL network more secure to use than a cable network, which is accessible to everyone on a network segment. ADSL has physical limitations like distance. However, potential problems may crop up between the customer sit e and the CO and quality and condition of copper wiring that can affect performance.

ADSL operates over a distance of up to 18,000 feet of 24-gauge copper pair wire in optimal conditions. According to specification, ADSL implemented over short distances will allow for full 9-Mbps speed depending on the line-coding. However, as the distance increases, line attenuation naturally degrades the signal, and less bandwidth is available for transmission. This effect is compounded when copper lines that have been in place for long periods of time are exposed to moisture and other environmental elements that can damage the lines. The telephone companies that have mixed very old and new technology in their build-outs over the years to accommodate a growing customer base will always be battling this ongoing problem.

Therefore, no site can be sure that it will get a predictable line speed appropriate to its distance from the CO until service is actually installed. Our tests of ADSL at MCI's labs (see "ADSL: Putting a Charge Into Your Copper Cable," May 1, page 139), however, show ADSL to be a die-hard technology that works well even with significant adversity from the combined elements of interference and distance, though potential problems may crop up as carriers begin to pair ADSL with other high-speed lines in large-scale carrier networks. ADSL may introduce interference on other lines carried in the same bundle, for instance. More than likely, however, the problem with ADSL will lie not in the technology, but in regulatory politics and business strategies of the telephone companies.

Like cable companies, local exchange carriers will have to struggle to provide a reliable data network infrastructure, as well as adequate management to handle increasing amounts of data and the rising number of subscribers. But unlike the cable industry, telcos must also compete against their own other product lines, such as ISDN and T1.

The fact is, local exchange carriers have spent plenty of money on ISDN, only to find it difficult, expensive and hard on the network. ADSL doesn't tax local switches the way ISDN does because it allows telcos to put voice traffic on their voice switches and move data traffic away from them. However, the local exchange carriers don't want to shell out more money to introduce and implement yet another technology. Nor do they want to introduce a service that will detract from their abundant crop of T1 subscribers.

This is what makes the new ADSL technology a crapshoot. Network designers trying to plan ahead have no way of knowing how the telcos are going to deploy or tariff the technology. The only reason telcos might ever kill T1 would be to drive demand for T3 services on the back end. As it turns out, they may be forced to provide inexpensive ADSL, or inexpensive T1, through competition from aggressive carriers that will soon invade their territory.

In the battle between ADSL and cable, most comparisons make the mistake of predicting that ADSL will cost much more than cable's HFC offering. These predictions state that an ADSL modem initially will cost up to $1,000, with ADSL access charges of up to $200 per month, plus additional usage fees. Often these figures are contrasted against the cable industry's initial cost of $100 for a cable modem, plus a flat monthly access fee of $35 to $60. The logical conclusion of such reports is that cable will become the WAN technology of choice by dint of its superior pricing scheme. But companies like Netspeed Communication offer ADSL equipment at prices of $100 per subscriber, and local carriers like Bell Atlantic have announced ADSL service starting at about $30 a month under certain conditions . So even in its early deployment phases, the ADSL offering is competing with cable on price.

In addition, nail-biting questions will arise with both ADSL and cable concerning how data will be routed to corporate networks and the Internet. What will be the option for Internet service providers (ISPs)? Will subscribers be able to choose among competing ISPs, or will the telephone carrier itself opt to be the ISP? Right now, most telephone and cable companies are too busy with technology issues to answer these questions.

Cable Access Technology: Playing Against The House
Cable networks, or CATV, were built using coaxial cable, which can carry an analog signal for distances spanning several tens of miles. To increase the distance, multiple signal amplifiers can be placed in the lines to reach destinations off the beaten path. This makes cable networks robust and less susceptible interference or cable wear occur that might otherwise degrade the signal. The network operates in a tree topology, where the local office (called the head end) receives the master signal from a satellite or other large fiber backbone and distributes it across the coaxial cable to several branches that eventually reach the end nodes (see "Hybrid Fiber-Coaxial Network Access," on page 55).

Over the years, cable companies have found that replacing coaxial cable with fiber lets the network span greater distances, carry more information and deliver a clearer signal without signal repeaters. Although some progress has been made toward replacing coaxial cable with fiber-optic cable, a majority of networks still have coaxial cable, and all use coax for the final span to the customer equipment. The resulting network is a hybrid fiber-coaxial network.

The construction of the HFC network lets cable companies break away from a traditional "broadcast-only" mode of operation over coaxial to offer two-way communication over unused bandwidth of the wire. This means existing cable eventually can pipe information to and from an ISP or corporate network. Of the 750 MHz of bandwidth available for a typical HFC cable network, only a small portion is used for traditional television program content.

The rest of that bandwidth is divided among data transmission, voice communication and other services. For upstream transmissions, the frequency range between 5 MHz and 42 MHz may be allocated to let end nodes communicate with the head end; downstream allocation is likely to range from 50 MHz to 750 MHz, depending on equipment. Common speeds of cable modems are expected to reach 3 MBPS upstream (from the customer premises to the head end) to 30 MBPS downstream (from the head end to the customer premises). Actual transmission-speed potential will hinge mostly on the transmission line-coding techniques the cable modems use and emerging interoperability standards. In both cable-based HFC networks and ADSL, access is expected to be billed at a flat rate; data is transmitted transparently to the CO or head end to be routed by the back-end equipment to backbone networks and ISPs.

What this boils down to is that cable companies will be able to compete with local telephone companies for voice services and Internet access. Cable companies say that since the network wiring is in place, piping up to 30 MBPS of data into a customer's location will be simple to implement and can be offered at an attractive price. However, there are limitations to the technology that will impede the cable industry's progress.

One limitation is the makeup of the cable network itself--a mix of coaxial and fiber. Until fiber can be deployed to the curb of a particular end node and upstream amplifiers are installed that allow two-way transmission, a customer will be able to receive the signal only via his or her one-way coaxial cable. As more coaxial lines are replaced with fiber, the cable network will continue to evolve into a more efficient, two-way, high-speed network.

Another drawback is that the HFC network is public, which means local users will be forced to contend for bandwidth in much the same way LAN users contend with one another for bandwidth. As with LANs, the number of cable subscribers on a given distribution network will have to be lowered, which, in turn, will raise the cost to the network service provider. Because users' traffic patterns tend to be bursty and sporadic, the problem isn't immediate, but as popularity grows and applications jack up bandwidth requirements, local access points run the risk of becoming congested.  

Security and WAN:

The topology of the cable network gives everyone access to the signal on a local loop--which means anyone with enough technical prowess can safely lurk in the privacy of his or her own lair without fear of detection. Encryption may eliminate some problems, but it increases the network cost, complexity and latency.

Network service providers and cable modem manufacturers recognize the importance of equipment interoperability. A standards initiative, called the Data Over Cable System Interface Specification (DOCSIS), aims to develop a set of interface standards to deal with the cable modem-to-customer premises equipment (CPE) interface (CMCI) between the head end and the subscriber equipment, operations support system interface (OSSI), telephone return interfaces (CMTRI) and the cable modem-to-RF interface (CMRFI). The success of this initiative will depend on the extent of participation between vendors and the organization leading the initiative, Multimedia Cable Network System (MCNS). This organization comprises major cable providers, such as Comcast Corp., Cox Communications and Time-Warner Cable, and it partners with Rogers Communications, Continental Cablevision and Cable Laboratories. The collective power of these cable companies to influence potential business provides a major incentive for any cable modem vendor to adhere to the standards.

MCNS' idea is to promote development of the cable system and equipment based on a common set of standards that will address connectivity issues between cable networks and customer sites; the CMCI will dictate the ground rules for interfacing the CPE with the head-end cable equipment. The cable modem termination system-network side interface (CMTS-NSI) will set rules for interfacing head-end-based equipment with backbone networks, server farms and management centers, which will exist behind the head end. The OSSI will focus on management of the equipment of the cable network equipment in terms of security, performance monitoring and accounting.

An interesting standard to watch will be the CMTRI, which deals with the interface between end-node subscriber equipment and the public switched telephone network (PSTN). CMTRI offers another means of upstream communication and lets cable companies compete with the local exchange carriers for voice telephone service. Perhaps the most important development within the DOCSIS will be the Data Over Cable Security System specification, the success of which will be the key to its future in business connectivity.

Other hurdles exist, too. Many network designers point to the cable industry's history of outages and poor customer service, noting that cable networks were built to broadcast entertainment and news, not to deliver critical data and transactions to hospitals during unfavorable weather conditions, for instance. Because the subscriber end nodes of a cable network are transparently piped into the head end through these modems, equipment such as Ethernet switches, hubs, management consoles and routers that relay information to the b backbone network and ISP’s must be installed and tightly managed by personnel at each head end. Many corporations insist that cable companies will never be able to manage these networks or provide 24x7 network resources.

The cable access is non-switched and dedicated, which means all calls must go through the head end, regardless of the destination of the data. To do this, the cable networks must be tied to ISP’s or be the ISP’s themselves. Close ties to third parties must be relied on to guarantee levels of service for Internet and corporate access. This may require a carefully knitted brigade of lawyers, contracts and management to avoid recursive finger pointing when increasing numbers of 3-Mbps to 30-Mbps pipes start flooding already congested ISP networks. Both the cable and telco industries will have a difficult time dealing with this situation.

Also under scrutiny is the financial position of the cable industry. Upgrading networks to HFC and building data communication infrastructures does not co me cheaply, and the cable industry's major source of revenue--supplying traditional TV content--is being gnawed at by the satellite communication industry, which is also planning to compete for data-access customers. There is also potential for telephone carriers to compete for the entertainment business through use of ADSL.

In fact, the pressures are so great that some cable providers have backed away from providing data service directly. Instead, they are hedging their bets by funding--and providing infrastructure for--start-up companies willing to take the risk. On the national level, service providers such as @Home Networks are partnering with local cable providers like TCI, Cox Communications and Comcast to help cable companies deliver a high-speed WAN infrastructure based on a dedicated backbone network. @Home Networks' corporate-level service, @Work, handles customers with higher demand for reliability and management. Other cable vendors are investing in this service or are starting similar services. Among the group are Time-Warner Cable, with its RoadRunner network, and Continental Cablevision, with Highway 1.

Place Your Bets
The hype is that ADSL and cable technologies will answer all challenges and replace traditional lines for local-loop connectivity. Both are designed to take advantage of existing telephone and cable network wiring to offer bandwidth on the order of 10 MBPS through dedicated pipes to the Internet and corporate data centers. They will boost productivity, pave the way for unlimited multimedia applications and Internet telephony/fax services, and save us tons of money.

The smart technology gambler will look beyond the marketing allure of a big, instant payoff and instead try to determine which technologies will truly meet tomorrow's demands, how each new technology is laid out, who the players are, what each new technology promises and what it can't guarantee.  


As enterprise WAN networking requirements evolve, Cisco continues to enhance the Cisco 7500 and 7200 routers, providing solutions that meet these requirements. New capabilities, such as multichannel networking, enhanced ATM interfaces, advanced software services, and platform enhancements enable enterprise network managers to efficiently utilize new and existing WAN services, reduce costs, and grow their WAN networks with the ever-increasing demands of enterprise network applications.

Frame Relay and ATM WAN technologies are growing at a rapid rate. New services like Frame Relay SVCs, Frame Relay/ATM Service Interworking, and FUNI are emerging to permit migration and integration of today's PVC-based Frame Relay landscape into the SVC- based ATM fabric of tomorrow. The Cisco IOS software provides industry-leading functionality in these areas to support all common industry standards. In addition, Cisco is committed to providing its customers with value-added implementations that give superior traffic prioritization, Internet access, security, cost-optimizing data compression, and integrated LAN/WAN software support. Cisco continues to build on these defining characteristics of the Cisco IOS software as Frame Relay and ATM WAN technologies permeate into networks worldwide. 


Topic Source:  & 

























































































































































































































































































































































































Home     Post an article    Suggestion   Contact
Disclaimer    Copyright; 1998-2002 Softron Systems, Inc. All rights reserved.
Home Discussion Forum Services Advertise Contact About Us Sign Up