As Ron Brown, Gigabit Ethernet program manager for Intel Corporation's Network Products Division (NPD), reviewed the final test results for NPD's Gigabit Ethernet standards design, he contemplated a critical decision regarding his new line of Gigabit Ethernet networking products. Ron was proud of his division’s fast and efficient development of the new line of products, but was worried about leveraging the innovation in the market and competing against similar technologies from other vendors. His most critical dilemma, however, concerned the lack of industry-wide technical standards for his product. Industry standards had been crucial to the success of earlier innovations. In light of this dilemma, Ron recognized two possible courses of action with respect to his Gigabit line of products:

1. Introduce NPD’s Gigabit Ethernet networking products before industry technical standards were established in an attempt to capture the immediate opportunities for growth, or
2. Delay release of products until standards were defined to ensure the products would be compatible with the newly devised standards and accepted by the industry.

His choice had major implications, not only on his Gigabit Ethernet product line, but also on the future viability of NPD as a successful business unit. While Ron worried that waiting for standards would delay his product launch and potentially allow competition to capture significant first mover advantages, he also realized that launching a product line before standards were established could prove very costly in the long run should the products prove incompatible. Moreover, Ron was concerned that either decision might affect Intel’s more lucrative microprocessor line. Brown thumbed through the latest market growth projections for Fast Ethernet as he pondered the implications of each option.

Inside Intel Network Products Division (NPD)

Intel, the world's leading manufacturer of computer microprocessors, created the Network Products Division in the mid-1980's to address the changing needs of business computing. As corporations moved away from traditional mainframe computing models toward personal computer (PC) networks, the demands on both PCs and networks increased dramatically (see Exhibit 2). Intel's decade of advancements in processor technology kept pace with corporate demands on PCs, however, network performance failed to follow suit.

By supporting a range of networking and network management products (see Exhibit 3), NPD not only addressed network performance, but also created an entire support system enabling corporate network managers to lower their costs of running and supporting computer networks. At the same time, however, competing firms who also produced network products gained increased acceptance and hence, market share. Intel's role in the networking community diminished, preventing them from meaningfully influencing industry network performance.

Consequently, NPD inherited a secondary strategy: utilize Intel's industry-wide influence and reputation as a neutral networking party to promote technical standards that would create the greatest increases in network performance, reliability, and value. Intel believed that greater network performance was essential to maintain growth in its core business of microprocessors. As detailed in Exhibit 2, most networks were running at the 10 megabits per second (Mbps), the same speed at which they had been running for over a decade. Consequently, the network had become the bottleneck of the PC. Slow networks prevented desktop users from realizing the full benefits of their newer and more expensive Pentium processors. Intel reasoned that if network speeds could be increased, thereby removing the bottleneck, it would clear the way for sales of faster microprocessors. Since the bulk of Intel’s profits came from margins on leading edge, or more advanced processors, using NPD to drive design and production of high speed networking products, and lower prices industry-wide, became a top priority for Intel.

Efforts to support this mission resulted in a blend of equity investments in networking start-ups, internal product proliferation and enhancement, alliances with significant networking vendors, and placement of key engineering staff on influential network task force boards -- all in support of microprocessor-selling network improvements.

Gigabit Ethernet

Gigabit Ethernet was the latest innovation in high-speed network technologies and was predicted to enjoy phenomenal growth rates and wide spread acceptance over the next several years. Gigabit Ethernet evolved from a slower networking language called Ethernet, the most popular method for connecting two or more computers (networking). Researchers at the University of Hawaii pioneered the technologies of a shared radio transmission channel to link communications devices as early as 1968. The Hawaii scientists termed their innovation "ALOHA," which existed as little more than an intellectual and technical curiosity until 1972, when Bob Metcalfe improved upon the ALOHA model. At the time, Mr. Metcalfe worked for Xerox Corporation at its now famous Palo Alto Research Center (PARC). Engineers and scientists at PARC are renown for innovations such as the world's first laser printer and Graphical User Interface (GUI).

Bob Metcalfe and his team of researchers were able to "speed up" the network and provide the capability to link over 100 computers. Their advances came in the form of Carrier Sense Multiple Access and Collision Detection (CSMA/CD), meaning that a networked computer would listen before transmitting its own data stream, thus allowing more than one computer to utilize the communication lines simultaneously and more efficiently. Metcalfe’s improvements, termed Ethernet, afforded him the ability to link up to 100 computers on the same (shared) bandwidth and pass information between computers at a rate of 10 Megabits per second (Mbps). Such improvements in bandwidth made computer networking practical for the first time. Today, bandwidth considerations continue to be the primary driver for improvements in network products (see Exhibit 4).

Gigabit Ethernet could transfer information from computer to computer at an alarming 1,000 Mega-bits per second (Mbps) (1 gigabit per second), thereby increasing the speed of a 10Mbps or 100Mbps network by a factor of 100 or 10 respectively. The growing use of high-speed local area network (LAN) technologies (see Exhibit 1) created a need for an even higher speed network technology in the systems beyond the desktop. Gigabit Ethernet used many of the same technical standards as previous Ethernet versions, and therefore, Intel believed, represented the most logical and appropriate solution for upgrading the network.

Intel and other networking vendors believed that Gigabit Ethernet had two advantages over competing technologies. First, upgrading to Gigabit from previous Ethernet networks was relatively easy. Current users would be able to upgrade their Ethernet networks at reasonable cost without interrupting existing service. Secondly, Ethernet was historically an open technology. The Institute of Electronic and Electrical Engineers (IEEE) established standards for Ethernet, thereby permitting numerous vendors to market and distribute the technology, making it the standard for LANs.

Intel believed these two advantages would make Gigabit the dominant standard for network backbones, just as 10 and 100 Mbps Ethernet had become the standard for local area networks. However, one major obstacle remained. IEEE representatives had failed to agree upon a standard for the new technology. Industry expectations were that a standard would not be agreed upon for at least 18 months. This complicated Intel's strategy of when to launch its first Gigabit Ethernet Product.

The IEEE Standards Process

The IEEE was an independent body created in 1969 by a number of industry vendors and academics whose purpose was to review and recommend computer and networking design standards. The IEEE's intentions were to simplify consumer choices and hopefully discourage vendor monopolies by creating open computer standards (computing methods that are not proprietary and work with a vast array of products and technologies). To participate in the IEEE process, companies submitted product designs or recommendations to the IEEE for review with hopes of approval and designation as the official industry standard. Unlike patent protection, the IEEE process did not safeguard designs. However, designs that gained IEEE approval created a significant advantage for the originating firm. Often, these firms increased revenues through licensing agreements to other firms that did not want to incur the development expenses necessary to support the standard.

The IEEE attempts to set standards supporting three major objectives:

1. Support product standards that are most compatible with existing and projected technologies in the retail sector.
2. Enable the highest performance within the constraints of the established technology.
3. Officially designate and endorse a design to not only encourage vendors to develop products to the accepted standard, but also persuade users to choose standardized products when making network and computer purchasing decisions.

Eventually, the IEEE was considered the "endorsing" body for computer and networking product designs. It wielded substantial weight with the buying public, and in many cases, consumers waited for IEEE approval before purchasing products, attempting to ensure that the maximum number of complementary products would be designed to operate seamlessly within the chosen standard. As a result, vendors also considered it in their best interests to delay product releases until IEEE standards were established.

However, due to its board's composition, the IEEE's objectivity was often suspect. Because of its prestige and a lack of qualified individuals willing to dedicate resources to their cause, the IEEE board was often composed of industry leading engineers and scientists employed by major networking companies. Therefore, many vendors felt that the firm with the largest board membership influenced standards approval. Also inherent in its objectives was the board’s task of supporting those standards most compatible with current and future industry products; once again supporting the largest installed base or the current market leaders.

Due to the questionable objectivity of the IEEE and the prolonged approval process (specifications commonly took 12 to 24 months), vendors often attempted to circumvent the IEEE standards process by releasing products to the marketplace in hopes that the market would choose its own standard, even though these products lacked IEEE support. Vendors made this "risky" attempt to bypass the IEEE by using one of two methods. The first method was performed by newly entering or industry-lagging corporations seeking to "leapfrog" the more established product leaders. Often these firms would design or license a technology they felt was vastly superior to current product offerings and introduce it immediately, hoping the market would recognize its obvious advantages and rush to purchase. As a result, the technology would become the industry "standard" and eventually win IEEE approval based on its large installed base, popularity, and developed compatibility.

The second method used to by-pass the prolonged IEEE standards process was usually performed by market-leading firms. Occasionally, despite offering technologically inferior products, dominant firms would be first to market and could achieve immediate acceptance by the shear weight of the firm’s reputation, influence, and channel strength. As a result, the IEEE approval process was often a formality that simply followed market acceptance of the design.

In addition to gaining immediate market share through early product introduction (pre-empting IEEE standards), firms were able to protect their proprietary product designs longer by avoiding IEEE design submittal and scrutiny by competitors; designs were always published for review. Smaller firms with limited marketing resources and minimal IEEE influence considered product design protection and continual product innovation as the most sustainable business strategies in the networking markets. As a result, these firms also relied most heavily on pre-empting IEEE standards.

Origins of Ethernet Standards

In the late 1970's, dozens of LAN technologies had emerged, Ethernet being only one of them. Moreover, superior networking technologies existed, and in some cases were supported by large, multinational companies such as IBM and AT&T. Ironically, it was not technical superiority that would make Ethernet the dominant networking protocol. Instead, it was the developer's drive to establish IEEE standards to govern the technology. In 1979, Xerox, DEC, and Intel lobbied the IEEE to create standards for Ethernet technology and the IEEE responded by setting guidelines. According to IEEE 801.2, the regulation that governs Ethernet standards, Ethernet would run at 10 Mbps per second and would utilize Carrier Sense Multiple Access and Collision Detection (CSMA/CD) to maximize network speed and performance. Such standards allowed as many companies as possible to produce and distribute hardware based on Ethernet technology, thus establishing Ethernet as the standard technology for Local Area Networks. By 1996, 83% of all installed network connections were Ethernet (IDC, 1996) and Mr. Metcalfe’s 3Com Corporation, the first commercial Ethernet enterprise, dominated the market.

Over the last 20 years, the exponential growth in the PC market and the development and popularity of software that demanded higher bandwidth mandated that networking companies market products that provided increased speeds. Technologists argued that networks using only a 10Mbps Ethernet connection would bottleneck, thereby inhibiting the PC's performance. The race for faster networks had begun.

Ethernet's First Upgrade - the Fast Ethernet Alliance

Ron Brown analyzed documents detailing Intel's successful support of the Fast Ethernet standard. Fast Ethernet was the first major improvement made to Ethernet since its commercial inception nearly a decade earlier. In effect, Fast Ethernet increased the speed of the network by a factor of 10 and ran at 100 MB per second.

Illustrating the effects of market power and IEEE influence, Intel helped define the Fast Ethernet Alliance (FEA) in 1993. The FEA was a collaborative group formed to define open specifications for 100 Mbps Ethernet and was composed of several acclaimed engineers from the major computing companies. The group eventually earned industry support and the backing of over 60 networking companies for the Fast Ethernet standard. Subsequently, in 1995, NPD launched a Fast Ethernet product line with considerable success.

Intel Network Products Division believed that the introduction of a faster networking standard provided an opportunity to advance its network strategy over several dominant industry competitors such as 3Com, Cabletron, Cisco, and the newly formed Bay Networks. Therefore, Intel worked to become the champions of Fast Ethernet. Intel trained its sales people to preach the advantages of faster networks and allied with other vendors in order to be first to market with Fast Ethernet Interface Cards and Systems. Product managers wrote white papers and published books advocating the benefits of the new innovation. The result of their efforts was a leading market share in Fast Ethernet Hubs and a second place share of the Fast Ethernet Interface Card market (see Exhibit 5). This contrasted greatly with their late entry into the original 10MB Ethernet market in which they never caught up to competitors who entered earlier (see Exhibit 5).

Although sales of Fast Ethernet remained a small portion of NPD's overall revenues, the product launch received a significant amount of visibility in the industry and was instantly adopted by a number of high profile customers and vendors. As a result, the Fast Ethernet specification was predicted to become the dominant networking technology by the turn of the century.

NPD's Gigabit Strategy

Brown assessed that the key to Intel's success in Fast Ethernet was both its early market entry and the endorsement of an IEEE standard. However, that same strategy would be impossible to duplicate with Gigabit Ethernet. Standards had not yet been formulated and much confusion remained as to the exact specifications of the technology. For example, vendors disagreed as to what type of cabling Gigabit Ethernet should utilize. Intel strongly believed that its customer relations would suffer if it launched a product that did not comply with IEEE standards. However, Brown also knew it wouldn't be long before powerful vendors entered the market with competing Gigabit products. Intel was faced with a critical strategic choice: ignore standards to gain early market entry, or wait for standards and yield first mover advantages to its competitors.

As he reviewed Intel's recent Fast Ethernet success, Ron Brown evaluated both the Gigabit proposal and Intel's current position in the networking industry. The industry had changed since the days of the Fast Ethernet Alliance. Despite continued efforts to remain an influential third party, Intel’s influence weakened significantly due to the rise of networking firms such as Cisco and 3Com. Analyzing the IEEE board composition, Ron knew of only one Intel employee on the board. To complicate matters, the board was headed by the chief engineer of Intel’s most formidable competitor, Cisco Systems. Moreover, industry experts were convinced that Cisco would launch Gigabit Ethernet products within six months despite lacking IEEE approval. Ron was confident that his NPD engineers had designed a superior product, but was deeply concerned about Cisco's influence and impending launch. Ultimately, he thought that Intel's chances of receiving standards approval, in the face of Cisco and other competitors, were slim at best.

Brown also worried about competing technologies. The major telecommunications suppliers (Lucent, formerly AT&T, and Northern Telecom) advocated a comparable technology called Asynchronous Transfer Mode (ATM) for networks. If Gigabit Ethernet failed to quickly make its presence felt in the market, end-users and resellers would turn to ATM for their immediate needs. Intel had thousands of hours and millions of dollars invested in Ethernet. By waiting too long, Intel might not only lose market share, but also endanger the future viability of Gigabit as a successful commercial innovation.

Competition

A multidimensional competitive environment existed for Gigabit Ethernet. Along one dimension lay the networking industry firms of 3Com, Cisco and numerous Gigabit startups. A second dimension was driven by the competing technology of ATM. Cisco, due to its relationship with IEEE, found itself in a particularly dominant position. Packet Engines, a Gigabit startup, had taken an aggressive strategy for hardware deployment. Alternatively, those firms who had invested in ATM technology found themselves in a precarious position with risk filled potential.

Cisco had been a pioneer and was now the industry leader in routing and switching solutions. Since 1993, the firm had been at the center of routing and switching technologies required for Media Access Control (MAC). Cisco’s position as a dominant force on the IEEE board had the potential of influencing the final configuration of the Gigabit Ethernet standard. Curiously, Cisco had adopted a policy of "wait and see" as to what the IEEE standards board would decide. Although the firm claimed it would maintain this policy, in late 1997 Cisco showed signs of pre-standard deployment of an entire product suite.

Packet Engines, on the other hand, had aggressively marketed Gigabit Ethernet products since 1996. In essence, Packet Engines wanted to be first to market to enable users to sample Gigabit Ethernet products. Therefore, Packet Engines launched a low-sophistication, low-cost "Gigabit Starter Kit" bundling solution. Packet Engines hoped their technology would convince the market of the benefits of Gigabit Ethernet. Then, they would expand their product line to increase their total market appeal. In late 1997, the firm generated tremendous excitement at the industry tradeshow, NetWorld+Interop, through its release of $1000 per port bundled solutions and demonstrations of advanced gigabit switch technology.

ATM was once the most promising technology in high-speed networking. It provided the flexibility to handle voice and data concurrently without the latency problems of IP. Cisco and all the major telecommunications players had a stake in the technology. ATM, while much more flexible as a general networking solution, was far too costly to install on every desktop. Gigabit Ethernet, meanwhile, promised to do so at a low cost. However, ATM still showed promise as users were not convinced of the need for 1000 Mbps bandwidth at the desktop. Even the Gigabit Alliance promoted a gradual transition to its technology starting at the highest levels of the network, an area where ATM was cost efficient. If it could not be proven that gigabit bandwidth was needed throughout the network, ATM still had a place and potential.

Distribution Channels and Customers

Complicating matters further, Brown estimated the customer response to both options. Network Products Division targeted small to medium size businesses for its network products. Previous research and segmentation studies had highlighted networks between 50 and 1,000 nodes as the "sweet spot" for Intel networking products. To reach its target market, NPD sold its products via a two-tier distribution system typical for networking vendors. Although Intel used over 100 distributors, 85% of its networking sales came from two primary distributors. These distributors held inventory for the vendors and maintained sophisticated information systems that were linked to both the vendor and its network of local resellers.

Intel's distributors, in turn, sold to thousands of Value Added Resellers (VARs) across the country. Intel VARs performed two purposes for its customers. First, they were hardware suppliers. They had access to distribution channels and could obtain needed networking and computer hardware for their clients quickly and at efficient prices. Second, the VARs performed their clients’ installation, service, and support functions. They employed network engineers that designed, constructed, and maintained networks. Therefore, they maintained close relationships with end users. They consistently surveyed and profiled their client's networking needs and anticipated future demands.

In addition to the VARs, Intel also distributed its products using Direct Mail Resellers (DMR). DMRs distributed catalogs to customers and took orders over the telephone or Internet. Products would then be mailed directly to the end user. DMRs were able to offer customers lower prices and faster service. However, they performed very little technical support for the end user. Such resellers were gaining popularity and market share over the past two years and currently comprised approximately 11% of Intel's total networking sales.

Intel also maintained a sales force of approximately 45 people. The sales force spent a majority of its time with VARs in their respective territories. Sales force duties included educating the VARs about the attributes of Intel's products and generating leads of end-users for the local resellers.

Demand for Gigabit

By all accounts, Gigabit Ethernet performed up to technical standards, operating 1000 times faster than those networks of only 4 or 5 years ago. However, Intel was still concerned about the amount of end user demand for the innovation. Did the customers need a faster network? Were they willing to pay the significant upgrade costs to achieve better performance?

In 1996 and 1997, major technical market research firms published papers on the new innovation. They projected enormous market growth for Gigabit (see Exhibit 6).

Customers, however, did not always echo such optimism. Intel's sales force was hearing mixed responses from its resellers and end users. Customers expressed concern about the need for faster networks. Some small and medium sized networks were still content running their 10Mbps networks. Others were more concerned about costs. Some companies had just upgraded to Fast Ethernet technology. Now, the network vendors wanted them to upgrade once again. Furthermore, migration ability was a particular concern. Currently, Gigabit Ethernet could not run on existing cabling (used in Ethernet and Fast Ethernet). Therefore, migrating to Gigabit Ethernet would mean more than just the capital expenditures of the hardware. It would mean re-wiring an entire office, building, or campus. The cabling issue was foremost on the minds of the IEEE board members when establishing standards.

Financial Estimates

Brown believed the market growth projections provided in Exhibit 6. He was confident that early market entry would yield substantial market share gains. However, he was also sure that consumers would eventually be displeased with products that proved incompatible with subsequent products from Intel or other vendors.

To estimate the profits under both options, Ron looked at historical pricing lifecycles for Ethernet and other networking products (see Exhibit 7). He knew the initial launch of the Gigabit Ethernet product line would entail an initial cash outlay of approximately $170 million. However, Intel was currently cash rich on its expected annual revenues of $26 billion and profits of $7 billion for fiscal year 1997, due to the success of its microprocessor line of products.

Performance Measures and Organizational Concerns

Ron's concerns about Intel's inability to influence the IEEE were compounded by NPD's divisional objectives and performance. Although NPD had a secondary mission of increasing network performance in support of Intel's microprocessors, NPD was still evaluated on profitability. In achieving only $500M in 1996 sales, NPD's costs virtually matched its income. Addressing upper management, Ron had repeatedly raised the inequity of his division's pursuit of IEEE standards as an irretrievable cost. Unfortunately, the division president's most recent comment represented majority thinking regarding NPD, "Pursuing network standards is in our own best interests. We have industry-leading technology and should be able to profit handsomely".

In addition to management's concern over division profitability, Ron had recently received several transfer requests from the division's top engineers. Several engineers cited the lack of year-end bonuses as a main reason for the transfer requests. Since year-end bonuses were dependent primarily on divisional profits, NPD's bonuses had fallen short of most other Intel divisional bonuses by a considerable margin. Additional complaints concerned rumors that NPD was going to support the IEEE's approval of Cisco's Ethernet standard and discard the superior design work internal engineers had developed.

Despite its mere 2% of overall corporate revenues, NPD commanded the attention of the Intel's senior executives. Frustrated with his options, Ron sent an e-mail to Intel CEO, Dr. Andy Grove illustrating his concerns. Grove's email reply simply stated:

Most of Ron Brown's colleagues at Intel's Portland, Oregon facility had already departed for the day. Ron understood that a timely decision was imperative. The distribution channels needed to know where Intel stood on the issue in order to plan for future client needs. Industry analysts also waited with anticipation for Intel to take a position on Gigabit Ethernet. As Ron folded up the market reports and logged off his computer, he felt confident that he had enough information on which to base his choice. He would announce and justify his decision first thing tomorrow morning.

Exhibit 1

Local and Wide Area Networks (LANs and WANs)

• Local Area Network (LAN): An assemblage of computer hardware and software that enables computers to share data, software, and other resources. The computers within a LAN are separated by small distances (1,000 meters or less).
• Wide Area Network (WAN): Provides computer communications over large distances for numerous computers. WANs often tie together groups of LANs, seperated by large geographic distances .

Definitions:

• Interface Card: Computer hardware that is installed in a workstation and used to connect the PC to a hub in order to form a LAN.
• Hub: Hardware used to connect from 2 to 120 computers on a LAN.
• Switch: Hardware used to connect hubs of PCs together. Allows more bandwidth than a hub and can be used to connect two or more LANs for form larger networks or a network backbone.
• Router: Network device that routes packets of information from the LAN to the WAN.
• Server: A powerful computer on the network that performs a service for the network such as file storage, print management or application hosting.

Exhibit 2

The Network As The Computer

While at the onset of the information revolution the central processing unit (CPU) was thought to be "the computer", today a very different environment exists. The modern paradigm is that "the entire network is the computer". This displaces the traditional bottleneck of the CPU to other parts of the system. The following diagram is intended to show that although today’s central processing units (CPUs) may run at speeds in excess of 500mhz or half a billion cycles per second, communication or data transfer rates with other parts of the system may be orders of magnitude less:

This results in the displacement of system bottlenecks:

Exhibit 3

Intel Network Products

Network Management Products

*LANDesk® Virus Protect

LANDesk® Management Suite

LANDesk® Workgroup Manager

LANDesk® Server Manager Pro

LANDesk® Configuration Manager

Integrated Technologies

High Bandwidth Networking Products

NetPortExpress™ PRO Print Servers

NetPortExpress™ PRO/100 Print Servers

Express Ethernet and Fast Ethernet Switches

Express 10/100 Stackable Hubs

Express Routers

EtherExpress™ PRO/100 Adapters

EtherExpress™ PRO/100 Mobile Adapters

EtherExpress™ Pro/10+ Adapter

EtherExpress™ Pro/10+ LAN Adapter

LANDesk® Network Manager

Intel Express Routers with VPN

Intel TokenExpress™ LAN Adapter Family

Intel Device View for Web

OEM Networking Building Blocks

Intel Fast Ethernet Network Adapters

Intel Fast Ethernet Networking Silicon

Intel Express Hubs, Switches, and Routers (OEM)

LANDesk® Client Manager

Network Complementary/Exploiting Technologies

Intel Internet Video Phone

Intel Intercast™ Technology

Intel Business Video Conferencing with ProShare® Technology

Intel Video Phone with ProShare® Technology

*Intel purchased LANDesk® in 1991 to provide network management software. Most products are managed by a newly created division, Systems Management Division (SMD) but are considered part of the NPD initiative.

Exhibit 4

What is bandwidth?

Bandwidth is an informal measurement for how much information a network can handle at any given time. Network bandwidth refers to both the amount of information that can travel through a portion of the network and the speed at which it travels.  

What factors influence bandwidth?

Two variables influence the bandwidth of a network. First is the communication line or cable. Advanced wiring, such as fiber-optic cabling, can handle far more information than conventional cabling such as coaxial or category five. However, cables that offer high bandwidth are far more costly than lower performing cables and, therefore, are used only in the network backbone where tremendous bandwidth is required. 

The second variable influencing bandwidth is the particular network hardware used in the network. Ethernet interface cards, hubs, and switches are examples of such hardware. The hardware controls both the amount of information that travels through the network as well as the speed at which it travels.  

Speed:

Network speed is measured in Megabits per second (Mbps). Whereas Ethernet cards, hubs, and switches send packets of information through the network at 10 Mbps, Fast Ethernet Cards, hubs, and switches are capable of sending packets through the network at 100 Mbps and Gigabit Ethernet sets the speed at 1,000 Mbps.  

Amount of Information:

Network hardware also controls the amount of information flowing over a single network line at any given time. Ethernet allows many different computers to utilize the same communications line simultaneously in what is termed "shared" network technology. The network hardware manages the bandwidth to ensure that each computer using the shared bandwidth receives the appropriate amount of network capacity. 

Shared Ethernet can be best visualized as a highway where many cars in many lanes (cables) merge into a single lane. Hardware controls which packet (or automobile) gets to go first. Furthermore, the network hardware will determine how fast the packets travel (speed limit) to their destination.

The growth of corporate networks and today's high bandwidth applications mean more and more traffic congesting a fixed amount of bandwidth in the network. When the network hardware is incapable of keeping up with the increasing amount of traffic, network performance diminishes (traffic jam). Fast Ethernet addresses this problem by increasing the speed limit of the network.

Exhibit 5

Fast Ethernet Market Share (percent)

Ethernet (10Mb) Market Share (percent)

Exhibit 6

Source: IDC #12382, November 1996.

Exhibit 7

Price Point Sampling

Back to the Menu