itmWEB: Switching in the LAN Backbone


..information technology management..

white paper

Switching in the LAN Backbone

An Anixter Technology/Business White Paper

Contents

Introduction

Information is the key to success. How an organization delivers information to its users, and in many cases to its customers, is a determining factor for overall productivity. The timely delivery of data can make or break a business. How well does your network operate? Is your LAN constantly experiencing unscheduled downtime? Do users experience slow or unacceptable response times from file servers? Is the cost of supporting and maintaining your network getting out of hand? Many organizations are grappling with these very real problems, but a solution does exist: Switching, backbone switching in particular, can provide a solid foundation upon which to expand a network.

Switching is catching on like wildfire. It is a logical transition for the evolving network. Most switches operate similarly to bridges in that they forward frames according to the destination MAC (media access control) address. However, frame and cell switches are much faster and use temporary virtual connections that link source to destination port. These dedicated connections provide more available bandwidth to network users than do shared network devices. Many companies are installing switches to alleviate existing network congestion and improve response times to and from application servers. Others find that switches are a lower-cost solution than are shared networking devices like bridges, hubs and routers and they provide many of the same functions. However, cost and performance are not the only concerns of network administrators. Long-term growth capability, flexibility, minimal disruption to users, preservation of current investments and support for multimedia applications are important too.

Figure 1 shows this evolution. Switches are used on floor levels to alleviate backup to department servers. A switch replaces the router in the backbone and that router is reconfigured to support traffic from the WAN.


Figure 1: The evolution to switching as a backbone solution

Figure 1: The evolution to switching as a backbone solution

Research firms are optimistic about the overall growth of the switching market. While sales for shared network devices continue to increase, albeit at a slower rate than initially, sales of switching products are exploding. According to figures provided by International Data Corporation (IDC), switched networking sales will equal or exceed worldwide sales revenue for shared networking products by 1999 at about $10 billion.

In the past many switching products have been sold for workgroup environments, typically to ease bandwidth shortages. These shortages manifest themselves in various ways, such as slow response times to and from application servers, crashing those servers and stifling certain segments. Workgroup switching, while alleviating local problems, only moves the traffic congestion to the backbone where more serious network failures occur. This shifting of the congestion is merely a temporary solution and introduces new challenges that must be addressed as network users increase in number, workstations improve processing capability, and network-centric applications such as interactive multimedia become more the norm. Applications that integrate voice, video, imaging and data will bend--if not break--shared backbone networks. Video and imaging files will drain bandwidth and the time-sensitive properties required for voice over LANs and WANs need special provisioning, such as quality of service (QoS) guarantees.

The need to move beyond shared-backbone networks is exemplified by the Internet. The Internet has supported some of these multimedia applications to a certain degree, but it has been experiencing slowdowns of late. This situation will only worsen as these traffic types continue to be transported across it. With the explosive commercial growth of World Wide Web applications, private networks are seeing a tremendous increase in Internet and intranet traffic. Some leading industry analysts estimate that about half of all the traffic on the Internet is a result of Web "surfing."

These same analysts believe that the Internet is on the verge of collapse, due largely to the bottleneck created by the router-based backbone design. Analysts believe this limitation, along with the increased use of multimedia applications, will be the Internet's downfall unless a technology like switching is implemented. Some companies are considering ATM cell switching as a long-term solution to ease the burden imposed by such demanding applications. ATM cell switching is capable of providing the additional QoS guarantees that are critical for time-sensitive applications.

The purpose of this white paper is to discuss the issues and concerns that pertain to transitioning client/server-based LANs from shared legacy backbones to a switched backbone design. In particular, we will examine some of the compelling reasons that justify such a move, from both a technical and a business perspective. The paper is organized as follows:

High-speed and Backbone Technologies

A network manager's job is not an easy one. With the plethora of high speed networking options available, determining which is the most appropriate can be mind-boggling. Some networks require different technologies in certain areas, while others have application requirements that dictate the strategic use of a particular technology. Whatever the case may be, prior to making a purchasing decision, certain fundamental concerns must be addressed about any LAN technology targeted for the backbone.

This section provides a brief summary of some of the most obvious choices for high-speed technologies and their capabilities when deployed in a backbone environment. Customers need to be armed with an understanding of the differences to help them first choose the correct backbone technology and then choose the right backbone devices for their networks.

100BASE-T (Fast Ethernet)

Operating speed available: 100 Mbps half duplex; 200 Mbps full duplex

Media access method: CSMA/CD (Carrier Sense Multiple Access with Collision Detection)

Architecture: Shared or switched

Topology: Star

Cable: UTP Category 3, 4 and 5; STP and fiber

Distance: Copper-100 m; fiber-2 km

Devices supported: Adapters, hubs, routers, analyzers and switches

Latency: Variable

The IEEE's extension to the 802.3u specification, commonly known as the Fast Ethernet standard, is very appealing as a high-speed solution, but will it stand the test of time? Many Ethernet devices support Nway.* But these devices fail to handle flow-control problems, although several proprietary schemes, such as backpressure, do exist. Unfortunately, time-sensitive traffic in this environment cannot be prioritized. Although proper network design can help reduce variable delays by limiting a single user per port on the switch, it is still impossible to distinguish the difference between traffic flows. As a result, there is no way to allocate buffers appropriately. Finally, Category 5 copper cabling has a 100 Mbps network diameter length limit of 210 meters. Although this limitation can be overcome by installing 100BASE-T devices other than repeaters to extend this distance, the limitation may hinder the effectiveness of 100BASE-T as a backbone technology. Due to its high-speed capabilities, 100BASE-T is excellent as a conduit to get to the backbone.

*Nway is part of the standard that defines autosensing between 10/100 speeds and the negotiation of a connection between devices at the highest-possible rate.

100VG-AnyLAN

Operating speed available: 100 Mbps half duplex

Media access method: Demand priority

Architecture: Shared or switched

Topology: Star and ring

Cable: UTP Category 3, 4, and 5 (requires all 4-pairs); STP; fiber

Distance: Copper-100 m; fiber-2 km

Devices supported: Adapters, hubs, routers and switches

Latency: Variable

A fierce competitor of Fast Ethernet is 100VG-AnyLAN, defined by IEEE in the 802.12
specification. Unlike 100BASE-T, it doesn't support fill duplex operation. 100VG-AnyLAN supports Token Ring and Ethernet frame types, although not in the same device. Translation between them is accomplished via a bridge or router; however, no products currently support Token Ring source routing. Since it is not based on collisions, 100VG-AnyLAN is better able to support a larger number of end systems. Distance is not as much as a limiting factor as it is in 100BASE-T with copper, network diameter over Category 3 is 500 m and 750 m over Category 5. Demand priority is the mechanism that prioritizes time-sensitive traffic. Like ATM, however, the100VG-AnyLAN applications must be rewritten to take advantage of this and to become QoS aware. Only a few vendors manufacture 100VG-AnyLAN switches for the backbone environment.

Isochronous Ethernet

Operating speed available: 16.144 Mbps half duplex

Media access method: CSMA/CD

Architecture: Shared or switched

Topology: Star

Cable: UTP Category 3, 4 and 5; STP

Distance: 100 m

Devices supported: Adapters and hubs

Latency: Variable in shared and fixed in switched implementations

Isochronous Ethernet, also known as IsoEnet, was developed by the 802.9 committee of the IEEE standards body. The 16.144 Mbps of bandwidth is divided up into a 10 Mbps channel for Ethernet data and 96 ISDN B-channels of 64 kbps for transmitting time-sensitive digitized voice and video. Another single 64 kbps D-channel handles ISDN (Integrated Services Digital Network) signaling control. ISDN capabilities allow seamless connectivity between similar end systems across the wide area. Copper is the only media type supported by IsoEnet and 16.144 Mbps is not enough bandwidth to seriously consider its use in backbones. Moreover, no switches are available at this time.

FDDI/FDDI II

Operating speed available: 100 Mbps half duplex; 200 Mbps full duplex (nonstandard)

Media access method: FDDI-token-passing

Architecture: Shared or switched

Topology: Star and ring

Cable: UTP Category 5, STP, fiber

Distance: Copper-100 m; fiber-2 km

Devices supported: FDDI-adapters, routers, hubs, switches and analyzers; FDDI II-adapters and hubs

Latency: FDDI-variable; FDDI II-fixed

The ANSI X3T9.5 committee ratified FDDI in 1988 and FDDI II in 1994. The most common FDDI implementation handles asynchronous traffic in a dual-ring, fault-tolerant design in which time-sensitive applications might suffer. However, synchronous FDDI has built-in priority mechanisms to overcome this problem by adapting its token-passing scheme. FDDI II divides the 100 Mbps bandwidth into 16 6.144-Mbps channels, each of which is further subdivided into 96 64-kbps channels that use ISDN protocols. FDDI II adapters have two MACs in order to multiplex packet and time-sensitive data to the same physical port. FDDI's fault-tolerant capabilities (dual rings, stations and homing), support a wide range of media types. High-bandwidth and widespread industry use make it well-suited for the backbone. On the downside, standards for source routing over FDDI need to be addressed. Because it was originally designed as a shared-media technology, FDDI has limited effectiveness. Therefore, many vendors are providing FDDI switches that will improve its longevity and use as a backbone solution.

HIPPI

Operating speed available: 800 Mbps half duplex; 1.6 Gbps half duplex--both may be full duplex

Media access method: Signaling

Architecture: Switched

Topology: Star

Cable: STP, fiber

Distance: Copper-25 m; multimode-2 km; single-mode-10 km

Devices supported: Adapters, gateways, switches, routers, storage and channel devices

Latency: Fixed

HIPPI (high-performance parallel interface) was originally defined in 1988 by the ANSI X3T11 committee. Gigabit speed is the main value of this technology, which was developed primarily for supercomputing applications. Data rates like those that HIPPI provides do not make sense in the typical LAN environment, i.e. one in which the data rates are limited by PC-based bus architectures (a PCI bus can handle peak burst rates of 1 Gbps). HIPPI has a credit-based flow-control mechanism to handle traffic congestion, making it better suited for multimedia applications that strain the available bandwidth.

Some of HIPPI's weak points are in media support: Copper cabling is 50-pair STP and limits network diameter to 200 m. While fiber specifications exist and are in use, they aren't standardized. HIPPI lacks multicast/broadcast support, so to provide this functionality, work-arounds are used. For example, ports are manually configured to direct traffic to a specific port where a multicast/broadcast server is attached. Interconnecting with other LAN-based media-access technologies is supported as long as the applications are running the TCP/IP (Transmission Control Protocol/Internet Protocol) Internet protocol suite.
For high speeds over longer distances, HIPPI, like ATM, uses framing based on SONET (Synchronous Optical NETwork).

Fibre Channel

Operating speed available: 100, 200, 400, and 800 Mbps half duplex--all support full duplex

Media access method: Header determined

Architecture: Shared or switched

Topology: Ring or star

Cable: Fiber, coaxial

Distance: Multimode-175 m; single-mode-10 km; coax: 25 m

Devices supported: adapters, hubs, switches and storage and channel devices

Latency: Variable or fixed

The ANSI X3T11 technical committee developed Fibre Channel in 1990 in response to high input and output requirements for high-end hosts and peripherals. Five different service classes are defined, which can provide varying levels of QoS. Flow control similar to the sliding-window mechanism used in TCP allows Fibre Channel to successfully construct a network of devices operating at various speeds. One major disadvantage is that it does not support UTP (unshielded twisted pair), and the vast majority of structured cabling plants still use Category 3 UTP. Despite these shortcomings, Fibre Channel is garnering more support, not just as a niche technology, but as one capable of providing high-speed LAN interconnectivity.

Gigabit Ethernet

Operating speed available: 1,000 Mbps half duplex; 2,000 Mbps full duplex

Media access method: CSMA/CD

Architecture: Shared or switched

Topology: Star

Cable: UTP Category 3, 4 and 5; STP; fiber

Distance: Copper-5 to 100 m; fiber-500 to 2 km

Devices supported: Adapters and hubs (prestandard)

Latency: Variable

The Gigabit Ethernet Alliance and the IEEE 802.3z are involved in developing the criteria needed to make this technology a reality, even though the standard will not be ready until 1998. The Alliance's intention is to build on the knowledge and standards that already exist for Ethernet and provide continuity and interoperabilty between 10 Mbps, 100 Mbps and 1,000 Mbps. Fundamental problems must be addressed, such as the tradeoff between speed and distance. Ethernet is collision-based and as such the correct interaction of

packet size, propagation delay and clocking timers is required in order to work. In general, as the speed increases, the distance of copper cabling decreases. For instance, with Fast Ethernet's tenfold increase in speed over 10BASE-T came an accompanying decrease in the overall network diameter from 2,500 m to 200 m. Initial deployment will probably include Gigabit Ethernet adapters and uplinks from 100BASE-T devices on fiber optic cabling with hubs and switch products following shortly thereafter. This is an excellent solution for Ethernet LANs.

ATM

Operating speed available: 1.54, 6.3, 25.6, 44.7, 51.8, 100, 155.5 and 622 Mbps--all theoretically
full-duplex capable

Media access method: Signaling

Architecture: Switched

Topology: Star

Cable: UTP Category 3, 4 and 5; STP; fiber

Distance: Copper-100 m; fiber-2 km

Devices supported: Adapters, hubs, routers, switches, multiplexers and analyzers

Latency: Fixed

ATM promises to be the next great wave of networking technology. It is capable of providing scalable amounts of bandwidth, various QoS guarantees and seamless integration of LAN and WAN environments. The compromise in the size of the 53-byte cell with header allows for better handling of time-sensitive traffic like voice and video. ATM might not be the best solution for all applications, but it addresses the bulk of the concerns that plague most LAN environments now and in the foreseeable future.

The ATM Forum's technical committees have recently resolved key issues such as flow-control mechanisms, low-speed access and interoperability between vendor switches that will allow global deployment of ATM. Connections may be manually configured (in PVCs [permanent virtual circuits]) with a QoS guaranty, or brought up on demand (in SVCs [switched virtual circuits]), with no QoS. Native ATM applications are now being developed to provide switched service with QoS. These applications should drive ATM to the desktop environment.

Figure 2: A rating of the technologies over various criteria

Figure 2: A rating of the technologies over various criteria 1 = lowest 3 = higest

The brief examination in Figure 2 of the high-speed and backbone technologies serves as a background for subsequent discussions about backbone switches. Customers should challenge the capabilities of the technologies they choose before they invest too much in those technologies. FDDI is the incumbent leader for fulfilling the needs of the backbone. However, ATM is becoming more widespread. 100BASE-T is sufficient for many applications and Gigabit Ethernet sounds promising but must be developed further before it can be taken seriously. Fibre Channel and HIPPI both have much going for them but are less appealing for client/server applications. The features and capabilities that should be inherent in a backbone technology include ample bandwidth, support for a wide range of cabling options that allow longer distances to be covered, mechanisms that handle time-sensitive traffic. These features, among others, will help ensure that your backbone solution will stand the test of time.

Backbone Switching

What is driving switching into the backbone? Resource centralization, server consolidation, aggregate bandwidth, reduction of network latency and LAN and WAN integration are some drivers. Backbone switches are capable of having servers, hubs, routers, and LAN segments collapse into them. The bottom line is that a backbone switch, however it is defined, must meet the needs of your backbone.

Backbone switches come in many varieties (Figure 3). Some support a single technology--for instance, only Ethernet; others support multiple technologies like Ethernet and Token Ring; while still others support shared, switched and serial technologies. This third type is sometimes referred to as an "enterprise switching hub" since it provides connectivity across these three domains. This section examines the distinctive features that a backbone switch should possess and why.

Figure 3: Examples of some backbone switches

Figure 3: Examples of some backbone switches

Backbone switches must first provide fault-tolerant capabilities--after all, this is the heart of the network. A modular chassis provides flexibility with hot-swap and redundant components. They should have very large address tables that can support numerous end-systems and segments. These switches need to have ample available bandwidth and the ability to pump it up as needed to accommodate future increases in the number of users and/or applications. In addition, backbone switches require large buffers and flow-control mechanisms in order to serve as an aggregation point for lower-speed segments. This is accomplished with a combination of hardware and software intelligence that allow more sophisticated filtering and routing than low-end switches. With router functionality and support for multiple technologies like Ethernet, Token Ring, FDDI and ATM "any to any" connectivity may be achieved.

Furthermore, even though VLANs (virtual LANs) are not widely implemented today, they will be soon. The 80/20 rule is changing. It is more likely that project collaborations are taking place remotely (outside the immediate workgroup) than ever before. Virtual LANs will facilitate the creation of "communities of interests" that transcend physical location. Communication among these VLANs may be handled within the unit and not require an external router. Other network management tools like RMON (remote monitor) and the impending RMON II specification provide traffic analysis capabilities for the switched
network.

Some other features to look for when evaluating backbone switches are:

Figure 4 shows examples of what a cross section from an enterprise switching hub might look like. The one on the left depicts a unit that employs a centralized processing module. In this instance, the segmentation and reassembly, or SAR, function that converts frames to cells takes place on an individual module in the chassis. This not only adds overhead to the function but the processing module thus becomes a single point of failure. The one on the right shows the distributed processing model and overcomes this limitation. Instead, the SAR function or layer 3 forwarding decisions can take place on the individual modules.

Figure 4: Examples of what a cross section from an enterprise switching hub might look like

Figure 4: Examples of what a cross section from an enterprise switching hub might look like

Many backbone switches are being designed with cell switching in mind. Make no mistake, manufacturers are doing this for a reason. It seems clear that cells, whether ATM or proprietary, are displacing frame-based internals.

Backbone Switching by Technology

In this section we will look at switches available for the backbone. These include products that provide Ethernet, Fast Ethernet, Token Ring, FDDI and ATM connectivity. We will not include Fibre Channel and HIPPI switches since our concentration is on the well-established client/server applications.

Switched Ethernet

Ethernet is based on a shared-medium approach--that is, all devices share the same wire to transmit data, which means one transmission at a time gets through. Switched 10 Mbps or full-duplex 20 Mbps (10 Mbps simultaneously in opposite directions) are available on most switches.

Media access method: CSMA/CD, a method analogous to a group conversation. You listen to ensure no one else is talking (if you are polite) before you attempt to speak to someone else in the group. If you start talking at the same time as someone else does, you both pause and then one of you starts again while the other waits.

Advantages

Disadvantages

The future for Ethernet switching is bright. Switched Ethernet provides about 70 percent of the revenue for the entire switched marketplace. According to IDC, the percentage of growth for the number of ports sold will be up 171 percent for this year over last. In 1995, the number of ports sold was about 2,000,000 and is predicted to surpass 5,300,000 ports in 1996.

Switched Token Ring

Like Ethernet, Token Ring is based on a shared-medium approach. Some Token Ring switches support half duplex at 16 Mbps and full duplex operation at 32 Mbps.

Media access method: Each device positioned on the logical ring waits its turn to examine the token to see if it must respond to the message being sent by another device or whether it can be the sending device. The token controls the transmission of data.

Advantages

Disadvantages

In 1995 the number of Token Ring switched ports was at about 39,000. The predicted numbers for 1996 should eclipse 1,000 percent growth to more than 500,000 ports.

Switched 100BASE-T (Fast Ethernet)

Like Ethernet and Token Ring, switched 100BASE-T, or Fast Ethernet, is based on a shared-medium approach. A major advantage of 100BASE-T is that it functions identically to 10BASE-T, but operates at 10 times the speed at a nominal cost increase. Some of these switches support full-duplex Fast Ethernet, which has a bidirectional aggregate speed of 200 Mbps.

Advantages

Disadvantages

IDC forecasts a 632 percent increase in the number of switched 100BASE-T ports-from 44,000 in 1995 to 322,000 in 1996.

Switched FDDI

FDDI was designed to be highly fault-tolerant, providing many larger networks with greater reliability. It also uses a token-passing scheme similar to that in Token Ring. This token passing may be suspended and provide a nonstandard full-duplex operation.

Advantages

Disadvantages

In 1995, FDDI switched ports numbered a little more than 60,000, but by the end of 1996 the number is projected to be about 171,000 ports.

Switched ATM

ATM is connection-oriented and provides QoS by guaranteeing certain traffic types higher priority. The delays for time-sensitive applications, like voice and video, are minimized because an ATM cell has a fixed length of 48 bytes and a 5-byte header (AAL5). With the recent completion of key specifications that provide flow control, signaling and routing between vendor switches a high degree of interoperabilty has been achieved. ATM is
delivering on its promises.

Advantages

Disadvantages

IDC numbers for ATM switched ports in 1995 were about 68,000, and for 1996 this number is predicted to increase 269 percent to 251,000 ports. Switched ATM provided about 10 percent of the revenue for the entire switched marketplace in 1995. This number is predicted to double by 1998, mostly affecting the switched Ethernet market.

Switches that support all technologies, with the flexibility to deliver a solution into any environment, have a marked advantage. Many manufacturers limit their options with products that switch only certain technologies such as Ethernet, which has a strong position in the market. Corporations that have already invested heavily in FDDI are likely to look to switches that support FDDI and ATM. Transitioning from FDDI to ATM in a cost-effective manner that protects the operational integrity of the network is critical for these companies. Failover from one technology to another, and vice versa, in the same unit is the most convenient solution; however, this could create a single point of failure.

Business Concerns

It would be naive not to consider some of the commercial implications of backbone switching. After all, the advantages that backbone switching provides for business might be all the justification required to explore and implement a solution. A major concern for most organizations is capital costs. With investments in the devices that typically constitute a shared network--routers, bridges, hubs and concentrators--many companies have shied away from implementing switched networks that include ATM backbones because of the impression that costs are too high. The following scenario (Figures 5 and 6) shows to be untrue and illustrates the real costs.

Figure 5: A shared network and a switched backbone network

Figure 5: A shared network and a switched backbone network

Figure 5 shows a shared network and a switched backbone network. The shared design has 120 shared Ethernet workstations connecting to a FDDI server farm through 10 multiport repeaters (hubs) and a router. The switched design consists of the same shared peripheral devices; however, the repeaters connect to switched Ethernet ports on the backbone device. The ATM switch provides LAN Emulation that allows legacy applications to interoperate between Ethernet endstations and ATM direct attached servers.

Cost Benefit Case Study

Shared Switched
100 Mbps Aggregate Backbone Speed 5.2 Gbps
100 Mbps Maximum Server Access 310 Mbps
833 Kbps Avg. client to Server Bandwith 8.3 Mbps
$53k Network Equipment Cost $40k
$441.00 Cost per Station $333.00
$530.00 Cost per Megabit $400.00

Figure 6: Cost analysis calculated assuming 10% of user population contending for resources

As you can see, the difference in the backbone bandwidth and the access bandwidth to the backbone is dramatic. If all users were attempting to contact file servers simultaneously, the switched network would far and away outperform the shared. And the cost (not including adapters) per megabit of a switched system is significantly lower than that of a shared.

Maintenance

Besides capital expenses, other recurring operating costs exist, including installation and maintenance. For example, maintaining a router is complex. Whether the expertise comes from personnel within an organization or from an outsource provider, costs can add up quickly.

Future Growth

Network growth can be expensive and shouldn't be ignored. As more users and applications are added to a network, greater bandwidth and functionality are required to accommodate change without major disruption of service. A scalable backbone switch will alleviate growing pains and save money in the long run.

Day-to-day Operation

Backbone switching can also save money in day-to-day operations. For instance, by installing backbone switches and increasing the amount of available bandwidth to LAN users, more work may be completed in the same amount of time. This translates into a real cost savings.

Support

Once companies decide to implement backbone switching, they will need reliable maintenance and support. Maintenance and support programs from a competent source can be invaluable and save money. Network management and risk management--tools that help assess network needs--also contain costs. Companies that resell and provide services for a wide range of networking products are better prepared to handle the intricacies of your network than companies that don't.

A network design should mirror the competency level of on-hand technical expertise; otherwise, outsourcing should be evaluated as option. If outsourcing for technical expertise is required, choose someone who has the experience and knowledge required to work with the products in your network.

Some points to be addressed when considering backbone switching:

Finally, planning a network with backbone switching is critical. Companies must decide how to transition to switching. Following is a sample infrastructure analysis that includes cost-saving ways to reorganize your network.

Step 1. Create a list of network assets.

Step 2. Determine where costs are (i.e., personnel, network management and downtime).

Step 3. Devise a cost-cutting plan.

Step 4. Weigh the benefits against the costs and look to backbone switching as a solution.

Conclusion

The networking marketplace has embraced switching in the LAN. Network administrators who recognize the key benefits of switching in the backbone--consolidation and centralization with fault-tolerant scalable products--are preparing to make important purchases of backbone switching products.

We have examined high-speed technologies and found that while some provide ample bandwidth, the correct mechanisms are not built-in to provide QoS for time-sensitive
traffic. Furthermore, increasing bandwidth does not make the problem go away, it only postpones the inevitable. Many companies are experiencing these problems as they try to move the bottleneck from the workgroup environment to the backbone. Backbone switching provides a solution to this problem.

We also pointed out certain characteristics that should be inherent in backbone switches. The list of characteristics is by no means all-inclusive, but it covers the bulk of the requirements. Although most backbone switches support low-speed technologies like Ethernet and Token Ring, they should provide connectivity to higher-speed backbone technologies like FDDI and ATM. We introduced a scenario to illustrate that solutions do exist, and they make financial sense too.

Copyright © 1997 Anixter Inc. All Rights Reserved.

Used by Permission.

Link to Anixter Inc.

Return to Spotlight Archive



The itmWEB Site™, Copyright © 2006, itmWEB Media Corporation,
All Rights Reserved - webmaster@itmweb.com