Moreover, the spectacular growth of the Internet is a global phenomenon. Between 1993 and 1996, the number of Internet hosts in Europe increased by about 600 percent, according to figures developed by the International Telecommunication Union. During that same period, the growth in Internet hosts in Africa and Asia was even more meteoric - about 839 percent for each continent.
This exponential growth will enable information access and electronic commerce in innovative ways that will have a fundamental, positive impact on how people live, learn and conduct business. Internet companies also are developing new methods of providing services such as telephony and videoconferencing. The ongoing growth of the Internet promises to have a far-reaching impact on overall network technology and evolution.
While the infrastructure of the Internet is being continually upgraded, new needs such as enhanced quality of service and demands for higher bandwidth access will significantly alter the development of the Internet. At this key juncture in the Internet's growth, industry leadership has a unique opportunity to identify and articulate an optimal framework that integrates emerging user needs and technology and industry trends with the most productive policy environment. Industry leadership in this realm involves not only pointing out the benefits of the Internet; it also includes articulating where the Internet is going and identifying and removing key bottlenecks to sustainable growth. And all this must be accomplished under the dual pressures of exponential user growth and increasing demand for sophisticated, bandwidth-hungry applications that involve features such as Java applets, electronic commerce, multimedia and graphics.
It is to achieve these goals that the Global Internet Project (GIP) - a voluntary, international organization of primarily Internet software, hardware and service providers and telecommunications industry CEOs and senior executives - presents this paper. It aims to identify a set of potentially crippling bottlenecks and offer forward-looking recommendations on how to eliminate or minimize such barriers to sustainable Internet growth. The approach we adopt is to segment Internet connectivity into a set of identifiable components and to discuss the bottlenecks associated with each. The components are:
A fundamental premise of this approach is the recognition that bottlenecks in each of these components must be eliminated in parallel, or the overall capability to users will be only as good as the weakest link in the chain. With its membership consisting of companies that are among the global leaders in each of the four areas discussed, the GIP has a unique contribution to make in this critical task.
A pervasive theme in our description and analysis of the foundations of continued Internet growth is the overwhelming need for open access for both providers and end users coupled with flexibility and interoperability of systems, networks and backbone components. Openness and flexibility have been the twin pillars of Internet growth; now that sustaining that growth in useful ways has become a more complex challenge, openness and flexibility assume even greater significance.
As an international group, GIP members believe that Internet expansion will depend to a great extent on the ability of companies and consumers to obtain products and services in a secure, flexible, convenient and easy-to-use manner. A fundamental precept of the GIP is that because the Internet is a global medium, it is critical to address its challenges globally. The group is working with appropriate national and international bodies to find answers to a variety of difficult issues to assure the best possible future for all members of the Internet community.
Each participating company has a direct stake in Internet development; we compete against each other in the marketplace. Just the same, we all share a vision of the Internet's future.
In our 1996 report on the Emergence of a Networked World, the GIP talked about:
As more and more people use the Internet and demand increasingly sophisticated applications, the time has come to devote more attention to the network itself, to alleviate and prevent any bottlenecks from slowing down the Internet's rapid expansion. These new applications and some of the bottlenecks users are experiencing are discussed in the following pages.
The Global Internet Project recognizes that a more advanced infrastructure must be integrated with today's network to support what users are starting to demand . . . and to realize our vision of what the Internet can and should become.
The GIP's policy architecture can be thought of as a pyramid, with network infrastructure as the foundation and commerce at the top level.
This paper on opportunities and bottlenecks targets the infrastructure layer of the above model. It focuses on the physical and technological components of the Internet and provides a foundation for discussions of a range of issues in the evolution to broadband Internet applications.
At each level of the pyramid, the GIP is engaging in partnerships with other groups that have demonstrated significant expertise in each category. The GIP will help to amplify and promote the messages of these partners.
Over the last two years, the GIP has made a series of policy recommendations, including:
As the Clinton administration's recent report on The Emerging Digital Economy points out, "The Internet's pace of adoption eclipses all other technologies that preceded it." Radio was in existence 38 years before 50 million people tuned in; TV took 13 years to reach that benchmark. Sixteen years after the first PC kit came out, 50 million people were using one. Once it was opened to the general public, the Internet crossed that line in four years.
At the end of 1996, about 40 million people were connected to the Internet. At the end of 1997, roughly 70 million people were online. In terms of electronic commerce, some anecdotal examples are illustrative. To cite only one: from $100 million in sales in 1996, Cisco's Web site at the end of 1997 was generating revenue of $3.2 billion.
The Internet's impact on the economy and jobs, while difficult to measure, can only be described as staggering. A study commissioned by the GIP concluded that a significant portion of GDP growth in the U.S. was a direct result of the Internet revolution. We also estimate that more than 1 million jobs have been created worldwide as a result of the Internet's expansion.
Fueling this growth are new software applications and functionality such as Java applets; all forms of digital commerce, both from consumers and businesses; multimedia and video applications, such as using the Web for education, healthcare and training, and a virtually endless list of additional applications. In terms of electronic commerce alone, an Information Technology Association of America (ITAA) survey predicts that the business-to-business electronic commerce marketplace will double in the next six months.
The Internet 2 initiative, which we strongly support, focuses upon some of the applications of the future and what sort of infrastructure is necessary to exploit these possibilities. This initiative of the University Corporation for Advanced Internet Development is geared to creating a research network that is between 100 and 1,000 times faster than the Internet today. It is making R&D investments with the goal of creating networks that are not only faster, but also smarter. These networks will have the ability to support new applications for distance learning, real-time collaboration and telemedicine. Background on some of these projects can be found at http://www.internet2.org.
In the current environment, however, bottlenecks and barriers threaten growth in the Internet's software segment wherever heterogeneity and openness fail to exist. Homogeneity and proprietary standards generally hamper the growth of software and the Internet Economy itself.
Software has become ubiquitous. Our economy and society now depend on software to run critical information infrastructures and to render personal home pages. Software exists throughout all portions of the Internet - from the content of Web sites, to desktop and server machines, to non-PC devices (such as hand-held computers and cable set-top boxes), to firmware in chips and other hardware, to the intelligence inside routers and other networking equipment. As oil was to the Industrial Age, software is the driver of the Internet Economy.
It may be useful to consider what might be called "The Technology Stack of the Internet Economy," including four layers of software, to appreciate the dangers of bottlenecks in this key segment.
Second, misuse of standards represents a bottleneck; some industry players may not disclose technical standards information to other developers on equal terms or at all. As a result, software development is either stopped or rendered ineffective or more expensive. Sharing ideas and improving the software code produced by others provide the sort of dynamic, ever-evolving growth that the Internet needs to remain vibrant and full of change. Curbing software development diminishes innovation. Closed or misused standards thwart innovation and diminish software development.
Finally, where proprietary standards and systems closed to interoperability prevail, Internet growth is stifled. Locking users into one platform or technology prevents choice; when consumers lack the ability to choose, new software ideas and products cannot gain access to the market, inhibiting the competition that produces change and growth. If software growth is tied to the monolithic pace of change in a homogenous environment, the economy is imperiled. One clear example of this is the Minitel. While very successful in its day, the proprietary, centrally run online service of France does not give users new features and functionality to compare with what is offered by the Internet and personal computers. Indirectly, Minitel has stifled the native development of the Internet Economy in France as business, government and consumers have no native way to transition from Minitel to something else.
Recent progress in the development of cable modems capable of supporting multiple megabits per second, and the recent effort to develop a Universal DSL (Digital Subscriber Line) standard offer encouraging signs, but the actual deployment to date is extremely limited compared to the expectations of the industry a year or two ago. Part of the problem has been the lack of interoperability standards that gave rise to nearly 50 types of DSL implementations from different vendors. Many of these DSL approaches lack interoperability.
While there is significant demand for high bandwidth access, consumers for the most part currently are limited to Internet access at analog modem speeds . . . typically tens of kilobits per second. A number of wireline and wireless technologies are competing to provide high speed access to the Internet. We briefly review the technologies - and the challenges facing their wide deployment.
Two categories of wireline access are among the most promising alternative technologies to emerge. The first involves utilizing twisted pair copper access lines in combination with sophisticated electronics at the user and network ends to provide digital connections of up to eight megabits per second. Basic rate ISDN technology, which has been under development for more than a decade, provides over 100 kilobits per second on most copper access lines. But it has been slow to be deployed in the United States, and higher speed alternatives have emerged to surpass basic rate ISDN in bandwidth and to assume its best control features.
The second category of alternative wireline access is a family of Digital Subscriber Line (DSL) technologies, developed by nearly 50 vendors, that supports speeds of a few megabits per second over twisted pair copper lines. DSL technologies show significant promise for removing the access bottleneck on a wide scale, but their deployment has been limited thus far. Only several thousand DSL lines have been installed in the U.S. to date.
Cable TV in homes, which has become widespread, offers a broadband access medium. However, cable plant must be upgraded with fiber optics to provide reliable two-way transmission, and it must be interconnected with switching and routing devices to provide broadband Internet access rather than simply support "one-to-many" broadcast applications. About 100,000 cable modems are deployed in the U.S. today.
Satellite transmission and new fixed-wireless access using a number of technologies at various parts of the radio spectrum provide the other possibilities for high speed access. At this time, DSL technologies and cable modems show the greatest promise of providing widespread high speed, two-way access - with wireless technologies playing a complementary role.
The development of technologies by numerous vendors, along with their slow deployment, points to major issues that must be addressed. Notable barriers are the development of often incompatible standards for the technologies, and the difficulty of new players in breaking into what is often an access monopoly with little incentive to rapidly upgrade technologies and offer new services.
An environment must be created where entrepreneurs have the maximum opportunity to deploy bandwidth-enhancing technologies such as DSL devices. Where there is an access monopoly, this can be accomplished by providing open and equitable access to both ends of the telephone loop for ISPs and other entrepreneurs, who then can deploy matching electronics at both the customer premises and network interface (see Figure 4). In the case of cable networks, the deployment of hybrid fiber-coaxial cable technology should be encouraged so that the embedded cable plant is capable of providing greater bandwidth in the user-to-network direction, where bandwidth is now constrained.
Continuing innovation makes this an addressable challenge from a technological viewpoint. Technological advances include the development and deployment of efficient wavelength division multiplexing (WDM) to provide greater fiber optic bandwidth, and the development and deployment of gigabit routers to support high throughput.
While the technology is available, businesses will innovate only in a climate where rational economic decisions can be made regarding investment. This requires an environment where multiple carriers will interconnect in a fully competitive market. What is needed is the ability of backbone providers to build facilities and interconnect with other backbone or access providers, bilaterally or multilaterally, under market-based terms and conditions.
The implementation of grades of service capabilities, along with varying levels of bandwidth commitments required for different grades of service, will allow recovery of investment in a way that reflects the true cost of providing each grade of service.
Caching and mirroring technologies, as well as advanced rerouting capabilities, are an essential part of optimizing backbone resources. Based on identifying often requested resources and available network paths at any given time, deployed bandwidth at any stage can be utilized more effectively with these network and server-based solutions.
The real challenge to server infrastructure, though, comes when Web sites move beyond the "first generation" style and start to become more dynamic: generating pages of information when they are requested, rather than in advance; personalizing the content of the Web site to match the requirements or preferences of a specific user, or handling real-time financial transactions, committed by a Transaction Processing system against back-end Enterprise operational databases.
When such sophisticated applications are coupled with increasing traffic, the demands on a server machine rapidly outstrip the capabilities of a single "box." Even the current generation of reasonably scaleable SMP (Symmetric multiprocessor) machines is no match for this level of both CPU (central processing unit) and I/O (input/output) capability. The bottleneck is simply described, but the potential solution involves this dilemma: maintain the impression of having a single, gigantic machine at the heart of a Web site, or use multiple machines, shatter the holistic illusion, and create a world where users are bounced constantly between the various servers which are hosting a particular Internet site.
It's all about "single-system image" - if your Web site doesn't give the impression that it is a single, self-contained entity, then marketing messages, branding and the appearance of core competence are quickly diluted. If, on the other hand, the site gives the appearance of a well oiled, single machine, coping effortlessly with anything that the Internet can throw at it, then the corporation behind that site instantly earns an aura of respect.
The other aspect of bottlenecks at the server end of operations is that of geographical location. While it is easy to picture an Internet where everyone is connected to everyone else with almost infinite bandwidth and low latency, in practice that is never likely to be the case. Different countries always will develop at different rates, and interconnections between countries will vary dramatically. This poses a challenge for online commerce. The whole point of going online is to reach a global audience, 24 hours a day, seven days a week, but for many potential Internet merchants, the connectivity path from the client to the server may be such that the site is effectively "closed for business" to those people. So now, the design of a server complex must encompass not only the ability to scale to a large size at a single site, but it also must offer the ability to reach customers in the far-flung corners of the network who may represent a huge untapped market . . . if only you were able to locate your server near enough to provide an acceptable quality of service.
Spreading the server capacity geographically would achieve two goals. It would allow reliable access from many parts of the world, and it also would provide a convenient parallelization and hence scaling of the servers. The challenge in this environment is the maintenance of integrity between the servers, both in terms of served content and the sharing of "transactional state" - the contents of your shopping basket might be a good analogy - between the geographically remote servers. Solving these issues will pave the way to removing the server bottlenecks of today's Internet.
The Internet's development to date has been as haphazard as it has been delightful, more a function of random entrepreneurial initiative than of mature planning. The time has come to balance entrepreneurial spirit with sophisticated planning, within the largely voluntary framework that always has characterized Internet growth.
The pivotal priorities of such voluntary management should be fostering and sustaining openness, flexibility and interoperability at all points of Internet access and interconnection. It would be incorrect and rash to predict that, absent these qualities, the Internet will come crashing down upon us. With or without voluntary collective intervention, the Internet will continue to grow, yet that growth may be limited and without the quantum leaps of service, quality and innovation that have marked Internet development to date.
Should proprietary interfaces, minimal competition and limited provider and consumer access dominate the next phase of Internet development, consider what might happen. Without the opportunity to build broad markets and recover and reward investors, potential Internet players may be discouraged from entering the arena, depriving it of vigor and dynamism. Content providers may hesitate to research, develop and introduce new products, fearing that distribution could be severely limited. This would be especially true if the planning of Internet growth excluded potentially promising global markets. End users would lack choice, and their expectations of a free flow of information could be frustrated. And if consumers ever become frustrated, all Internet players will feel the effects.
If, on the other hand, openness, flexibility and interoperability are nurtured and preserved, the reach of the Internet will continue to expand exponentially, providing new opportunities for end users and Internet companies alike. Markets will be virtually limitless, encouraging the development of new technologies and products. Enthusiasm and support for the Internet will remain strong, creating a fertile climate for innovation.
It is toward the achievement of the goals of openness, flexibility and interoperability that the Global Internet Project offers the following overall recommendation, as well as more specific recommendations in the four key segments discussed.
GIP would be pleased to work with other interested organizations in hosting, planning and supporting the Internet Foundations Summit, as it might be called. We see the summit as the launching point for comprehensive, innovative thought and dialogue about creating the best environment for Internet development that serves the most people in the most ways.
If you work in an office, chances are the computer or terminal on your desk is connected by wires to some or all of the computers and terminals used by your colleagues. You send written work and e-mail messages back and forth among yourselves this way, and you're linked as well to common equipment, such as printers and fax machines. In short, you work on a network.
Increasingly, chances are that your office network is connected to one or more standard telephone lines. By means of a modem, your network's computer files can be translated into digital messages and delivered over those lines to the computer "address" of another network - in a different city, country, or continent. That second network with which you're communicating similarly communicates with any number of other networks. And therefore, so can you. Congratulations. You're on the Internet.
In an important sense, it really is just that simple. The Internet is a "network of networks" - a collection of computers, software and stored data, all woven together into a telecommunications system that is the world's first digital information infrastructure. This simple idea, an easily accessible global library of knowledge joined to a vast array of electronic tools with which to use it, promises without exaggeration to revolutionize human life in the next century. Indeed, as we have seen, the Internet already has altered fundamentally the way we teach and learn, buy and sell, practice medicine, organize our workdays, publish, read, and entertain one another - all for the better. And these benefits are just the advance guard . . . provided the appropriate openness of the Internet can be maintained in its next great growth phase.
"We will think of this period as a dramatic time of transformation for telecommunications.... The first milestone has already occurred. And that is the introduction of the Web itself - making it easy to access a wide variety of people. I think all consumer devices, televisions, telephones, essentially the ordinary home computer devices that you think of today as having a special connection in one form or another, will all connect into your data communications concentrator which will be your PC. That PC will manage the retrieval of television programming or educational material or computer software or checking your bank account or retrieving anything else, any other kind of information from the Web. This is what I believe."
The Internet has no specific physical location. No particular institution or group owns or administers it. Its material content is not stored in a central location. With the exception of some administrative functions such as the allocation of domain names at the top level, there is virtually no central anything about the Internet. Instead, it is a deliberately decentralized, voluntary, unregulated, loosely structured, infinitely flexible process . . . a process by which almost every conceivable strand of knowledge, in every conceivable form - audio, video, and text alike - can converge at once, in an instant, at a given user's command.
To the uninitiated, of course, the enthusiasm of telecomputer veterans for the Internet, their understanding that it represents an achievement as unusual and history-making as the steam engine, can seem like so much vaguely mystical hype. But the enthusiasm is justified, as a glance at the history of the Internet makes clear.
ARPAnet was designed to preserve itself through redundancy. Each of its computers and networks was electronically linked to as many of the others as possible, so that communication among them might travel along innumerable routes on the system's map. No single linkage between any two computer sites was required for the maintenance of any other linkage.
The Internet was purposefully designed by Bob Kahn and Vint Cerf to interconnect ARPAnet with two other, very different, networks. To do this, ARPA during the period 1973-1978 developed the now-standard protocol called "TCP/IP," the programming mechanism by which networking computers break down "packets" of information for transmission to other computers, ensure that those packets get to the right place, and reassemble them in useable form once they do. This protocol was implemented on many operating systems and finally, in 1983, deployed across all ARPAnet hosts. By sheer accretion of networks, ARPAnet had become a largely nondefense system, and the Pentagon withdrew its proprietary components.
A high-speed network called NSFNet emerged in 1986 to build upon ARPAnet's accomplishments. NSFNet was designed to do two things:
"The Internet is like a 20-foot tidal wave coming, and we are in kayaks. It's been coming across the Pacific for thousands of miles and gaining momentum, and it's going to lift you and drop you. We're just a step away from the point when every computer is connected to every other computer, at least in the U.S., Japan, and Europe. It affects everybody - the computer industry, telecommunications, the media, chipmakers, and the software world."
Consider the simplest e-mail message you might send to a friend in another state. When you push the "send" button on your home computer, that message is transmitted by your computer's modem over standard analog telephone lines to, perhaps, a commercial online service (such as America OnLine) that provides Internet access. You may not subscribe to such a service; you may instead purchase Internet access from an Internet Service Provider (ISP). Or you may have Internet access by affiliation with one of the large corporations, universities or governments that maintain their own private network gateways. Or you may gain entrance through one of the growing number of nonprofit, library, association, or community groups that now offer free or reduced-charge Internet usage.
No matter what the initial step, software on your home computer places your machine on the Internet as all the millions of computers that are permanently connected. The software in your PC breaks the e-mail up into digital packets. Usually, your e-mail message is sent to a relay which takes care of sending the mail to your targets using TCP/IP. Those separate TCP/IP packets travel over wires and cables through a system of local and regional networks to the Internet "backbone," a global network of fiber optic cable and switching stations maintained by large telecommunications companies such as MCI and AT&T. Once this is accomplished, the process is reversed. The packets are reassembled, and your e-mail reaches its destination. Its parts have traveled no preestablished route. The computers involved automatically choose the best, the least busy, the most efficient cyberspace highways for your mail - and for every Internet transmission, no matter how large or complicated.
Microsoft Chairman Bill Gates has warned that if the "Internet Gold Rush" of organizations eager to "get their stuff on the "net" outpaces infrastructure improvements necessary to process that "stuff," then even the highest-capacity cables of the system's backbone will be clogged. And if system gridlock or slow user-end connections become a persistent problem, if the best information cannot be made available to users quickly and efficiently, then the Internet may lose public support and interest. It's a valuable caution.
"One thing is certain. Over the next decade, computer speeds will rise about a hundred-fold, while bandwidth increases a thousand-fold or more . . . The chief beneficiaries of all this invention . . . will be the people of the world, ascending to new pinnacles of prosperity in an Information Age . . . Communications bandwidth is not only the secret of electronic progress. It is also the heart of economic growth, stretching the webs of interconnection that extend the reach of markets and the realms of opportunity. Lavishing the exponential gains of networks, endowing old jobs with newly productive tools and unleashing creativity with increasingly fertile and targeted capital, the advance of the telecosm offers unprecedented hope to the masses of people whom the industrial revolution passed by."
There is compelling evidence that Gilder's vision will be realized with the aid of promising new tools. Digital "ISDN" service that replaces the analog modem - and transmits data at least five times more quickly, allowing clear video imagery over regular wires - is already widely available from American phone companies . . . and may already have been surpassed by Digital Subscriber Line (DSL) technologies, which support speeds of a few megabits per second over twisted pair copper lines.
In addition, coaxial wires strung for home cable television services, offering Internet connections 1,000 times faster than the average modem, are being deployed in growing numbers around the country. Data compression and packet switching technology, which squeezes digital information into smaller, more efficient chunks and moves it more efficiently across the Internet, is constantly being upgraded.
And perhaps the most promising bandwidth solution is tantalizingly close. Because of an explosion of infrastructure upgrades in the 1980s, American phone companies have installed 25 million miles of fiber optic cable to transmit digital information through glass at (almost literally) the speed of light. Fiber optic installation is continuing at a rate of 4,000 miles a day. A fiber optic cable no wider than a human hair can send data through at more than 1 trillion bits per second.
The speed of the Internet backbone rose 400-fold between 1987 and 1992. Overall Internet bandwidth doubles every year. Bill Gates himself has predicted: "We'll have infinite bandwidth in a decade's time."
"The Internet is mostly software. We talk about it as if it's hardware - personal computers, switches, routers and optical fiber, but . . . the heart of the Internet is the software - the applications and the protocols that make it go. Software is infinitely malleable and anybody who can think up a way to program something can go off and do that, and they do . . . people will simply do things that you wouldn't have normally thought of doing, especially graduate students who don't have to worry about profitability. So they will just go try things out and some of them will turn out to be just mind boggling, like the World Wide Web. So I anticipate big surprises and monumental changes in the way in which the Network [the Internet] can serve us, simply because of the fact that it's software, and software can do anything."
The Web now incorporates, and is integrated with, all the other basic tools of the Internet: the ability to connect remote users to the full computing power of other networks and systems; internetwork text and file transferring functions; person-to-person communications programs such as e-mail and back-and-forth typed "chat," and public bulletin-board "newsgroups" and subscription interest-group "listserv" applications. The Web is all these tools and strengths in a single, easy, spectacularly multimedia whole.
As the Internet gets faster, as bandwidth inexorably expands, as the day approaches when knowledge of every kind - in all its forms - will converge at the click of a button on the screen of any given computer user in the world, then thinking and its industry no longer will be limited by place or time, but only by the interest and curiosity of the human mind. The Internet is not just a scientific revolution. It is the greatest and most consequential scientific revolution in history. And it is happening in your lifetime.
Note to users: All information provided is of a general
nature and is not intended to address the circumstances of any particular
individual or entity. Although we endeavor to provide accurate and timely
information, there can be no guarantee that such information is accurate as of
the date it is received or that it will continue to be accurate in the future.
No one should act upon such information without appropriate professional advice
after a thorough examination of the facts of the particular situation.
1998 GIP (Global Internet Project) All rights reserved.