The Exabyte Internet

An Analysis of the Growing Flow
of Billions of Gigabytes of
Digital Data and Its Impact
on Public Policy for the Internet.

May 1, 2007
US Internet Industry Association

Author: David P. McClure

Publication Date: May 1, 2007

Published by: US Internet Industry Association
1800 Diagonal Road
Suite 600
Alexandria, VA 22314
(703) 647-7440 Voice
(703) 647-6009 Fax
(703) 851-4784 Mobile
Info.3@usiia.org
http://www.usiia.org

Formed in 1994, the US Internet Industry Association is the primary trade association for companies engaged in Internet commerce, content and connectivity. USIIA serves its members through legislative advocacy and professional services. The association is headquartered in Washington, DC.

David P. McClure is President and Chief Executive Officer of the US Internet Industry Association. A technologist by education and experience, McClure has held positions in the Internet, computing, aerospace and environmental services industries. He is widely published on technical and business topics, and is the author of more than 35 white papers related to Internet and broadband policy, governance and economics.

© Copyright 2007, US Internet Industry Association. All rights reserved.

Table of Contents

Executive Summary

Introduction

Defining an Exabyte

Opportunities qf an Exabyte Internet

Concerns About an Exabyte Internet

Policy Implications for the Exabyte Internet

Footnotes

Executive Summary

Weiji [way-jhee], modern Chinese for "crisis"

Whether or not the coming growth in data on the Internet constitutes a looming crisis may depend on the definition of "crisis." As expressed by an American translation of "weiji:"

"The word "crisis" is composed of two characters.
One represents danger, and the other represents opportunity.⁽¹⁾

Recent research reports have suggested that the growth of data created by humans, some of which must inevitably transit the Internet, may create a situation in which it becomes impossible for the Internet infrastructure to handle the capacity demands for data, particularly video. While some have challenged the precise numbers, it is clear that the amount of data traffic on the Internet is growing at an accelerating rate. This will soon create a situation in which billions of gigabytes of data ­ called exabytes ­ will transit the Internet each year.

This is the Exabyte Internet.

And while individual consumers may not reach a level of data transfer of this magnitude in the short term, the Internet in the aggregate will. It will be necessary to continue high levels of investment in both the Internet backbone and in the backhaul subsystems that link each consumer to the Internet.

The Exabyte Internet promises to bring new services, applications and opportunities to both consumers and businesses in America, but it will not do so automatically. In addition to investments in additional infrastructure and capacity, there will need to be innovations in core technologies, innovations in network management, flexible strategic plans and a national broadband policy that supports investment and innovation over regulation and a continuation of the status quo.

This paper examines the available data relating to both the danger and the opportunity: the concerns as we move toward an Internet transporting billions of gigabytes of data; the likely effects on consumers, businesses and network operators; and the ramifications on Internet public policy as we grow from a Megabyte Internet to an Exabyte Internet.

It suggests four policy platforms necessary to the future of the Exabyte Internet:

• A better understanding of the data.
• A more realistic view of the Internet infrastructure.
• A "light hand" regulatory approach.
• A focus on investment as the national priority.

Introduction

When Discovery Institute Fellow Bret Swanson noted in a Wall Street Journal editorial that a coming "exaflood" of data could exceed available bandwidth capacity of the Internet⁽²⁾, he gave focus to growing concern that the Internet might be at risk of failure.

Swanson's research was not new -- other researchers and analysts have likewise charted the growth of data in the early years of the 21^st Century. In 2003, researchers at the University of California ­ Berkeley estimated that all of the data created from the dawn of humanity to that date amounted to 5 exabytes, or 37,000 times the total information stored in the Library of Congress.⁽³⁾

Other studies indicated that even this enormous amount of data is being surpassed every single year by the growth of technologies for data creation, distribution, transit and storage.

The global research firm International Data Corporation released a study in early 2007 entitled, "The Expanding Digital Universe: A Forecast of Worldwide Information Growth Through 2010⁽⁴⁾," indicating that in 2006 some 161 exabytes of digital information were created and copied. That amount of information is equal to three million times all the books ever written. According to IDC, the amount of information created and copied in 2010 will surge more than six fold to 988 exabytes -- a compound annual growth rate of 57%.

Similar concerns appeared in other research. 'Telecommunications Trends: TMT Trends 2007'⁽⁵⁾ from the Deloitte Touche Tohmatsu Technology, Media and Telecommunications Industry Group concluded that, "One of the key possibilities for 2007 is that the Internet could be approaching its capacity. The twin trends causing this are an explosion in demand, largely fueled by the growth in video traffic, and the lack of investment in new, functioning capacity. Bottlenecks are likely to become apparent in some of the Internet's backbones, the terabit-capable pipes exchanging traffic between continents. Investment, either in laying new cable or lighting existing fiber, may be stifled by continuing falls in wholesale capacity prices. Similar capacity constraints may well appear in the ISP and telecommunications networks that provide broadband connectivity to consumers. The impact may be most noticeable in the form of falling quality of service."

A January 18, 2007, research report from analysis firm UBS stated, "The growth in online video has significant implications for bandwidth demand on the Internet. Standard-definition video requires roughly 450 times the bandwidth per second of a web-browsing session. Taking it a step further, high definition video requires 2,700 times the bandwidth per second of a web-browsing session. Based on these numbers, Level 3 and other participants in the industry have suggested that existing Internet backbone capacity is insufficient to transport high-quality commercial video."

Nor is this strictly a U.S.-focused issue. Netcom CEO Bill Barney has noted that the total intra-Asia Internet usage between 1998 and 2005 grew by 13,857 percent. Barney forecasts that at the rate that demand for high-bandwidth applications is growing in the region, the 15,360 Gbps of capacity currently serving Asia will be completely used up by 2012.⁽⁶⁾

Though there was some quibbling over the methodology used in each of these reports, and of the final numbers derived to quantify the size of the growth in data, there is consensus that the amount of data created ­ data that will ultimately be transmitted across the Internet -- will grow at a sufficient pace to warrant serious consideration.

Still, the numbers do not yet mean an Internet in imminent danger of collapse. In 1995, at the dawn of the modern Internet, the total amount of data that traversed the Internet backbone was 1.5 million Gigabytes⁽⁷⁾. By 2006, this had grown to 700 million Gigabytes⁽⁸⁾. While this is significant growth in Internet traffic, it still represents only 0.5% of the 161 exabytes of data that were created.

Whatever the exact dimensions of the exaflood, it will impel us to significantly enhance network capabilities to maximize the promised benefits of new and innovative applications.

Weiji [way-jhee], modern Chinese for "crisis"

Whether or not the coming growth in data on the Internet constitutes a looming crisis may depend on the definition of "crisis." As expressed by an American translation of "weiji:"

"The word "crisis" is composed of two characters.
One represents danger, and the other represents opportunity.⁽⁹⁾

Swanson himself noted: "I don't equate the exaflood with a crisis. I think it is a great sign of advance. It will be good for hardware, software, network, and applications companies. It will be good for consumers. It will happen if we let it.⁽¹⁰⁾"

And while individual consumers may not reach a level of data transfer of this magnitude in the short term, the Internet in the aggregate will. It will be necessary to continue high levels of investment in both the Internet backbone and in the backhaul subsystems that link each consumer to the Internet.

This paper examines the available data relating to both the danger and the opportunity: the concerns as we move toward an Internet transporting billions of gigabytes of data; the likely effects on consumers, businesses and network operators; and the ramifications on Internet public policy as we grow from a Megabyte Internet to an Exabyte Internet.

Defining An Exabyte

To get a sense of the discussion, it's helpful to define the units of measure ­ the bytes ­ that make up the digital content world. A single digital character, a letter or number, is a single byte. A typewritten page is about 2,000 bytes, or two kilobytes, and a small, low-resolution image is about 100,000 bytes, or 100 kilobytes. There are about 5 million bytes, or five megabytes, in the complete works of Shakespeare, and a pickup truck full of books might amount to one billion bytes, or a gigabyte. One billion of those book-filled pickup trucks, or one billion gigabytes, is an Exabyte.

Opportunities of An Exabyte Internet

The opportunities for new services, new applications and enhancements in lifestyles are apparent in both the recent growth of the Internet and in plans already underway for the next generation of services.

For consumers, this should mean more Internet options at a sequentially lower cost. Yahoo has announced that it will allow its mail system users to store an unlimited amount of data on its servers for an unlimited amount of time.⁽¹¹⁾ Wal-Mart, Netflix and Amazon have all announced plans to transmit DVDs online. A growing number of universities are making course lectures available on-line, often in real time. And telemedicine programs are transmitting medical images and linking patients with distant specialists for real time consultations. Added to this are new and pending applications for IPTV, video on demand, security, personal communications and telecommuting.

The impact on businesses also should be a net positive. While requirements for local data storage and longer-term data archiving will increase, the costs associated with this storage will decrease overall because of the ability to replace file cabinets and paper storage with less-costly electronic data storage. Data storage itself will become cheaper, more efficient and more secure.

Businesses small and large will experience an increased ability to place content on the Internet for use by their current and potential customers. Shopping online will become more effective through use of better tools for web site hosting and management, the ability to provide more documentation and support services to customers, and enhanced options for higher quality of service in content delivery.

At the same time, businesses will experience cost reductions in core services ranging from accounting programs and customer relationship management to business intelligence, and supply-chain collaboration among suppliers and buyers, as these services are taken online and delivered to the desktop in real time.

Both consumers and businesses will need to make informed and effective decisions about the capacity, services and support they will need for the applications they choose to use. While some may have a lesser need for video and audio, for example, others will experience stronger requirements for security, communication, server hosting, advertising and data archiving.

Realizing these opportunities, however, will be virtually impossible using only the Internet structure that exists today. The flood of data, applications and services will demand innovations in network management, strong and increasing investments, and effective business models that have been tested in the marketplace.

As noted by David Farber, the Carnegie Mellon engineer considered to be the "Grandfather of the Internet:" "An updated Internet could offer a wide range of new and improved services to provide these services, both the architecture of the Internet and the business models through which services are delivered will probably have to change.⁽¹²⁾"

Concerns About An Exabyte Internet

Network operating companies are pushing hard to deploy and expand the broadband infrastructure of the United States. According to the National Telecommunications and Information Agency (NTIA), a unit of the US Department of Commerce, America leads the world in the number of broadband subscribers, with 64.6 million as of June, 2006⁽¹³⁾.

Between June, 2001 and June 2006, the number of homes with broadband in the United States increased by 599%, from 9.2 million high-speed lines to 64.6 million. During the same period, satellite and wireless broadband grew by 5,998%. As of June 30, 2006, DSL connections were available to 79% of households with telephone service available, and cable modem service was available to 93% of households with cable service available⁽¹⁴⁾.

The primary concern about the amount of data being generated is that an increasing proportion of it must eventually traverse the Internet, if only on its way to a storage facility. And while that Internet traffic today is only a tiny percentage of the total, this inexorable migration of information and the growth in use of broadband by both businesses and consumers raises concerns about the ultimate ability of the thousands of networks that make up the Internet to build infrastructure quickly enough to keep pace.

There are four specific concerns:

1. Growth in the amount of Internet traffic will continue to accelerate. As more data is created, the human desire for communication will drive an increase in the amount that is communicated across the Internet via email, instant messaging, file sharing and other technologies. According to network infrastructure company CacheLogic, more than 60 percent of Internet traffic is being taken up by peer-to-peer swaps, and about 60 percent of those swaps involve video content⁽¹⁵⁾. This doesn't even count legitimate video and music downloading traffic. This surging traffic is driving records sales of routers worldwide as network operators struggle to keep pace.⁽¹⁶⁾ Worse yet, current video traffic is at best in HDTV format, which is already obsolete. The newer, "4K" video images have roughly 4,000 horizontal pixels ­ offering approximately four times the resolution of the most widely-used HD television format, and 24 times that of a standard broadcast TV signal.
2. The number of data streams will increase substantially in the coming decade due to the transition from IPV4 (Internet Protocol Version 4) to IPV6. Internet Protocol version 6 is a network layer protocol that will replace IPV4, the current version of the Internet Protocol, for general use on the Internet. The main improvement brought by IPv6 is the increase in the number of addresses available for networked devices. IPv4 supports 232 (about 4.3 billion) addresses, which is inadequate for giving even one address to every living person, let alone supporting embedded and portable devices. IPv6, however, supports 2128 addresses; this is approximately 51028 addresses for each of the roughly 6.5 billion people alive today. With such a large address space available, IPv6 nodes can have as many universally scoped addresses as they desire⁽¹⁷⁾. Unfortunately, each of these addresses, if attached to a person or device, will begin generating its own stream of data ­ even if only to acknowledge its existence on the Internet.
3. The evolving business models of internetworking will continue to shift traffic from the Internet backbone to a peered system in which content is streamed directly to consumers. This shift substantially alters the dynamics of infrastructure growth in that it drives a need to invest not only on the Internet backbones but also on the more expensive buildout of the "last mile" connection to consumers.
4. There may be insufficient investment in Internet infrastructure to accommodate the growth in data because government policies inhibit or interfere with the ability of networking companies to attract investment capital from the marketplace. National and state policies toward broadband innovation and deployment directly affect the ability of network operating companies to raise investment funding in the capital markets. To the extent that these policies inject uncertainty into the marketplace, threaten the potential return on infrastructure investments, or interfere with the ability to operate the networks in a way that maximizes capacity and efficiency, such policies would directly and negatively limit the ability of network operating companies to support expanding data traffic on the Internet.

Network operators are today investing heavily in infrastructure, both in terms of capacity and in innovations to better manage the flow of traffic across the Internet. According to CMP Technology's research unit Heavy Reading, worldwide sales of long-haul DWDM (dense wavelength division multiplexing) products grew by more than 30 percent in 2006, to $1.8 billion, as network operators scrambled to add capacity in the wake of surging demand for bandwidth.⁽¹⁸⁾

Some have suggested that simply adding capacity will resolve the issues of explosive Internet traffic growth. Though additional capacity is an important part of the overall strategy, it will not in itself resolve all of these issues, for three reasons:

• The Internet is not a single entity. For the sake of simplicity, it is easy to think of the Internet as a single, integrated network. In reality, it is a loose confederation of thousands of private networks that have agreed to use the same protocols. There is not even a single "Internet backbone," but instead dozens of backbones operated by companies of every size. Simply "adding capacity" means coordinating the funding, logistics, rights of way, purchases and installations of thousands of updates in virtually the same time frame ­ a nearly impossible task.
• The problem is at the fringes, not the center. Companies that carry the highest volume of Internet traffic are capable of managing their growth, given sufficient investment capital. It is at the outer edges of the Internet, where users connect, that the biggest problems will be experienced. This last-mile loop of connectivity is already struggling to finalize the deployment of basic broadband to all of the interested users, many of whom choose to live in remote geographic areas that are expensive and difficult to reach with today's technologies. Others serve urban centers where available rights of way and existing fiber networks are already under duress, making them expensive and difficult to reach. For these companies, investment capital for capacity may be very difficult to acquire.
• Even at present growth rates, data will eventually out-grow capacity. Market research group Teleography notes that average traffic across the net increased 75 percent last year, while capacity grew 47 percent⁽¹⁹⁾ ­ a trend that is likely to continue, even as capacity increases. Internet traffic growth may be infinite; the capacity growth needed to absorb more traffic is not a given that will happen by itself.

In addition to investment in infrastructure and capacity, network operation companies will need four other critical assets as they move forward:

• Scalable core technologies capable of faster, more effective throughput. Though we have been able to extract ever-greater efficiencies in transmission of data over the Internet through compression and innovative routing, some core technologies will simply not offer sufficient throughput to meet the demands of the future.
• Innovative technologies and solutions for network traffic management. In an age in which unsolicited spam comprises 80 percent of the world's email volume, it is not enough to simply make more capacity available. Continued management to route around bottle necks, efficiently meet applications' technical requirements, and ensure that delay-sensitive packets get through will remain key to smooth operations. Efficiency techniques, such as those already in use by major e-commerce vendors to ensure the delivery of their content through companies such as Akamai, will be necessary to ensure that priority content gets where it needs to go without transmission delays that disrupt performance.
• Flexible, prescient strategic plans. Juggling growing data streams, changing consumer needs and tastes, rapidly-evolving technologies and the uncertainties of the capital markets will place strains on even the best of network operating companies. The ones that are most successful in meeting the demands of their customers will be those with a clear vision, strong strategic plans, effective business models and the ability to understand and react positively to emerging trends.
• National, state and local policies that stimulate continued investment in broadband networks. Estimates of what it will cost to build networks capable of meeting an ever-increasing flow of data ­ and then deliver that capability to every American business and household ­ range from $300 billion⁽²⁰⁾ to as much as $1 trillion. Public policy should first focus on enabling network operating companies to secure and utilize the investment capital needed to meet these estimates and provide consumers with affordable broadband services. Etablishing a pro-growth environment that fosters stability and profitability in the broadband markets will better enable America to meet its national goal of universal broadband.

Policy Implications For the Exabyte Internet

This latter point speaks to the heart of the policy implications for the Exabyte Internet. Four policy platforms are needed to support the needs for broadband growth in America:

• A better understanding of the data. Limited views of Internet broadband growth and penetration, such as the annual OECD rankings and data based on single perspectives of Internet use are not particularly useful foundations for public policy, and may in fact lead to erroneous or unintended conclusions. More comprehensive approaches to broadband data collection, including the Notice of Proposed Rulemaking promulgated by the Federal Communications Commission in April, 2007⁽²¹⁾, will provide a more suitable policy foundation.
• A more realistic view of the Internet infrastructure. As noted, the Internet is not a single, monolithic entity. No single company or group of companies controls it or should, and no single company or group of companies can meet all of its needs. Just as no single set of Internet applications will benefit every consumer, no "one-size-fits-all" regulation will enable every broadband network to meet consumer needs in an efficient and affordable way.
• A "light hand" regulatory approach. The most successful changes in Internet laws of the past decade have been those that advanced simple, light-handed deregulation with a minimum of unintended consequences. This approach has left the Internet with a minimal tax burden, fewer legacy telecommunications regulations from the 19^th and 20^th Centuries and fewer barriers to growth. The result has been a vibrant, fast-growing and innovative Internet industry that is well-positioned to meet the demands of the future.
• A focus on investment as the national priority. The most pressing broadband need in America today is to craft an infrastructure capable of reaching every American business and household with the bandwidth and services need by those businesses and households. The goal of any national broadband policy must be to find innovative and effective ways to encourage the private investment to build that infrastructure.

End Notes

1. President John F. Kennedy, April 12, 1959

2. “The Coming Exaflood,” Bret Swanson, The Wall Street Journal, January 20, 2007

3. http://www2.sims.berkeley.edu/research/projects/how-much-info/datapowers.html

4. http://www.emc.com/about/destination/digital_universe/

5. http://www.deloitte.com/dtt/research/0,1015,sid%3D1012&cid%3D108298,00.html

6. “Asia to run out of bandwidth by 2012 - Asia Netcom” Total Telecom, April 25, 2007

7. University of Minnesota Digital Technology Center, as cited in the Chicago Tribune

8. University of Minnesota Digital Technology Center, as cited in the Chicago Tribune

9. President John F. Kennedy, April 12, 1959

10. “Exaflood Debate,” Bret Swanson, March 24, 2007, http://www.disco-tech.org/2007/03/exaflood_debate.html

11. http://www.techcrunch.com/2007/03/27/yahoo-mail-announces-unlimited-storage/

12. “Hold Off On Net Neutrality,” By David Farber and Michael Katz, Washington Post, Friday, January 19, 2007; Page A19

13. NTIA Fact Sheet, May, 2007, at http://www.ntia.doc.gov/ntiahome/press/2007/ICTleader_042407.html

14. FCC Data cited by NTIA, May, 2007, at http://www.ntia.doc.gov/ntiahome/press/2007/ICTleader_042407.html

15. http://news.com.com/Net+video+explosion+triggers+traffic+jam+worries/2100-1025_3-6042300.html

16. http://www.itnews.com.au/newsstory.aspx?CIaNID=46129

17. http://www.en.wikipedia.org/wiki/Ipv6

18. http://www.heavyreading.com

19. http://www.baselinemag.com/print_article2/0,1217,a=205078,00.asp

20. "National Imperative: Broadband Everywhere by 2010,", Cisco Systems, January 15, 2002

21. Notice of Proposed Rulemaking, Docket 07:37, at http://hraunfoss.fcc.gov/edocs_public/attachmatch/FCC-07-17A1.pdf