Current (year-end 2007) annual Internet traffic growth rates
U.S.	50-60%
World	50-60%

Year-end 2007 monthly Internet traffic estimate
U.S.	750-1250 PB (PetaByte = 1015 bytes)
World	3000-5000 PB (PetaByte = 1015 bytes)

Year-end 2007 estimates for monthly Internet traffic (GB per capita)
Australia	1.0
Western Europe	2.3
Japan	2.6
U.S.	3.0
Hong Kong	17.0
South Korea	17.0

Estimates for Australia and Hong Kong are based on official government statistics, while that of Japan is derived from cooperative ISP data collection in that country. In all cases, extrapolations were made to provide estimates for year-end 2007. Figures for other countries are based on snippets of information of varying degrees of reliability, as well as confidential reports by some service providers.

Traffic growth rates from publicly observed sites

See Data for more detailed information and Methodology for descriptions and definitions.

• Summary

How much traffic is there on the Internet, and how fast is it growing? Network technologies, architectures, and business models all depend upon and influence the answer. There are many other important measures of the health of the Internet, but traffic volume plays an especially important role.

The press is full of alarms about "exafloods" of traffic, primarily video, that might overwhelm the Internet. This is motivating calls for new business models, with many implications for issues such as "net neutrality" (see MW2007, And2007, MI2007, and McC2006). But there is very little solid data about what is happening on the network, and many conflicting estimates. As one striking example, at the end of 2005, John Chambers, the CEO of Cisco, claimed that Internet traffic was growing at about 100% per year Boslet2005, and similar claims are common (e.g., Roberts2006). Chambers also predicted both in 2005 and in a keynote at the NXTcomm conference in June 2007 Chambers2007, Duffy2007 that growth might accelerate towards 300 to 500% per year, and that the internal Cisco corporate network traffic load is currently growing at such rates. On the other hand, in August 2007, Cisco released a pair of white papers Cisco07a and Cisco07b which estimate that worldwide Internet traffic growth has been around 50% per year over the last few years, and project similar growth for the next few years. (See also the article DW2007 that draws on the Cisco white papers and other sources.) Thus even within a single company, and one that plays a central role in making the Internet function to boot, there are wide disparities in growth estimates, between 50 and 100% per year for current growth rates, and between 50 and 500% per year going forward. And there are some far more outlandish estimates floating around, including a Converge! Network Digest story that claims current Internet traffic is about 1,000 times as high as it actually is. So how can anyone make reasonable plans for the future? And in particular, how can we avoid on one hand a capacity crunch that strangles vital communications, and on the other hand another debacle like that of the turn of the century, when well over a hundred billion dollars was wasted in the United States alone building networks on the false assumption of "Internet traffic doubling every 100 days"?

Data traffic patterns have changed in the past, and may do so again in the future. Hence it is desirable to have a source of solid information about Internet traffic, to see if any dramatic changes are taking place.

• Purpose of this page and request for additional information

In order to shed some light on traffic trends on the Internet, this web page collects information about Internet traffic from a variety of sources. Most are public web sites (for which URLs are provided), over 100 at this point, that continuously monitor traffic on a variety of networks. They include the largest public Internet exchanges in the world, and together account for a substantial fraction of the world's Internet traffic.

Sites being monitored are listed on our Data page (also available through the menu bar at the top of this page), where by clicking on the link to each site one can obtain URLs for the home page of each site, and more directly for the traffic reporting page of that site.

Statistical analyses of growth trends derived from these sites are given, for individual years and for the full periods for which data is available.

Some evaluations on this page of trends also use anecdotal data and detailed statistics provided under non-disclosure by some carriers. Various official and semi-official statistics are also used, and references are provided.

Additional data and pointers to traffic statistics are solicited. Please write to odlyzko@umn.edu.

• MINTS monitor of publicly available traffic statistics

In spite of the widespread claims of continuing and even accelerating growth rates, Internet traffic growth appears to be decelerating. In the United States, there was a brief period of "Internet traffic doubling every 100 days" back in 1995-96, but already by 1997 growth subsided towards an approximate doubling every year CO1998, and more recently even that growth rate has declined towards 50-60% per year. South Korea, which along with Hong Kong appears to be the world champion in Internet traffic intensity, experienced its brief burst of "Internet traffic doubling every 100 days" around the year 2000, when broadband was widely deployed. It then appears to have had several years of annual traffic doubling, but currently (based on anecdotal evidence) is also growing at about 50% per year.

Traffic growth rates of 50% per year appear to only about offset technology advances, as transmission capacity available for a given price steadily increases. Thus although service providers are pushing to throttle customer traffic, an argument can be made that they should instead be encouraging more traffic and new applications, to fill the growing capacity of transmission links, Odl2008.

The current slowdown in growth rates may not be permanent. There are exabytes of data EMCIDC2007 that can conceivably lead to those famed "exafloods" that could strain the network. The key issue is how quickly (if at all) that data will move to the Internet Odl2003a. The main purpose of this site is to provide access to a variety of publicly available traffic monitors. Since few carriers or private networks report any statistics, the sample represented here is not fully representative of the Internet as a whole. There are also issues with the quality of the data that is available, and with the methodologies that we use. This is discussed in more detail on our methodology page. Hence the data presented here should be used with caution, and in particular should not be used to conclude anything about any particular network or exchange point. But MINTS does provide a snapshot of traffic trends at a wide variety of Internet sites, accounting for a substantial fraction of the world traffic. Hence any major changes, such as a sudden acceleration or deceleration in growth rates for Internet traffic, can be expected to be reflected in the data reported by MINTS.

The table below is a brief summary of our observations. More details, with pointers to the data sources, are on the MINTS Data page. That page provides not only statistical analyses for the sites that are monitored (over the full period for which data is available, or for particular years, as selected by the reader using the menu bar at top), but, perhaps most important, links to the original traffic information sources.


• Comparison to other studies and additional comments

The estimates for Internet traffic volumes presented on this page are roughly consistent to those in several other sources. For example, the recent Cisco white papers Cisco07a and Cisco07b have comparable estimates for 2006. Table 1 in Cisco07b gives world Internet traffic in 2006 as 2,303 PB/month (but without specifying whether this was average for the year, or some specific time of year), whereas the MINTS estimate is 2,000 to 3,000 PB/month for end of that year. An important warning is that the graphs and wording in Cisco07a are about general IP traffic (thus including closed networks that are not truly part of the Internet, but use IP, the Internet Protocol, such as the IPTV services of various telecom firms). The background white paper Cisco07b has detailed data and discussion that make this distinction clear. The estimates on this site are all for Internet backbones only. Thus they ignore IPTV traffic, as well as the data flows on private switched wavelengths, for example.

A Nov. 2007 study, "The Internet Singularity, Delayed: Why Limits in Internet Capacity Will Stifle Innovation on the Web" from Nemertes Research, that has attracted extensive attention, and is available at report. It projects an annual doubling of Internet traffic over the next few years.

A Jan. 2008 report from the Discovery Institute by Bret Swanson and George Gilder, "Estimating the exaflood" projects a 54% annual growth rate in traffic out to 2015.

A March 2008 report commissioned by the Fibre to the Home Council Europe states that residential users with connections speeds of 100 Mbps or more consume three times as much data as typical ADSL users, and projects that home access speeds will continue to grow at about 50% per year. report (PDF).

Our estimates for intensity of Internet traffic (as measured in GB per capita) show the U.S. and Japan as about equal. On the other hand, there is reliable evidence that Japan has far faster residential access links Harden2007. This contrast must reflect different usage patterns. (It has been argued for a decade that the main driver for deployment of data networks is the desire for fast transaction latency, as in quick web page downloads, database updates, or video transfers Odl1999. Hence a competition just in the speed of the link, of the type that appears to have developed in Japan, is not surprising.)

A more interesting case is that of Hong Kong and South Korea. Their access speeds appear to be comparable to those in Japan, but their traffic intensities are far higher, and South Korea, with about one sixth of U.S. population, has about as much traffic as all of North America

• Internet growth trends, Moore's laws, and related issues

The Internet is not an isolated phenomenon, but rather a part of the general Information and Communication Technologies (ICT) revolution. Data traffic growth is related to a variety of other growth rates, such as the famed Moore's Law for semiconductors. For a brief discussion of these laws, how many of them are changing, and what the implications might be, along with references, see: Data and speculation about Internet growth trends

• Other public and semi-public sources of traffic information
• A. Government statistics
• B. Cooperative industry data collection efforts
• C. Individual service provider reports
• D. Market research firm reports
• Links to other sites with relevant information about Internet

Traffic volumes are a very crude measure of the state of the Internet. Investment is driven by profits, and those have only a slight relation to traffic. (As just one example, mobile voice telephony has at least twice the number of users that broadband Internet does, far more revenues, and substantial profits, and, on balance, may very well have had a larger impact on the economy than the Internet. Yet mobile voice volumes are small.) Thus one should bear in mind many other measurement aspects of the Internet and other telecommunications services that are relevant. Here are links to some sites that are of interest (with more to come later):

• CAIDA, the Cooperative Association for Internet Data Analysis, provides a wide variety of detailed and scientifically validated studies of Internet traffic.
• PCH, Packet Clearing House, a source of extensive data about Internet exchanges and global network development.
• FCC, the U.S. Federal Communications Commission, publishes a range of statistics for U.S. telecommunications.
• OECD, the Organisation for Economic Co-Operation and Development, provides a portal with links to a variety of broadband studies.
• Invisible Hand Networks provides bandwidth market data.
• Renesys provides information on Internet connectivity.
• DSL Prime is a newsletter on broadband trends.
• Internet World Stats provides usage and population statistics.
• Akamai monitor of network quality (lacks hard data)
• Internet Traffic Report is another monitor of network quality.
• ipoque report presents studies on composition of traffic on some networks, with special attention paid to P2P traffic.

©2002 Regents of the University of Minnesota. All rights reserved. The University of Minnesota is an equal opportunity educator and employer. Last modified on August 30, 2007. Troubleseeing the text? | ContactU of M | Privacy