Sei sulla pagina 1di 36

insight from data / knowledge into action

Study to Assess Broadband Bandwidth Usage and Key Trends in Europe


February 2008

Produced independently by Ventura Team LLP for the Fibre to the Home Council Europe.

Executive Summary
In this study we have examined broadband traffic patterns over several years at the European level and for a sample of 100,000 true broadband homes (i.e. those providing consumer services of 100Mbit/s or faster) in Europe. Our first objective was to test Nielsens Law for Europe. In 1998 a US based web page usability consultant called Jakob Nielsen postulated a `Law of bandwidth that projected a 50% increase every year in the broadband speed available to a high end customer. This was similar in concept and formulation to Moores Law for the computer industry but slower (Moores Law implies 60% pa increase in computing power). We find that Nielsens Law on average has held remarkably well for more than 20 years although Government intervention in Sweden lead that country to greatly exceeded the rate of progress predicted by the Law. Using this Law as a guide we expect 100 Mbit/s services for high end consumers in the France in 2008, Poland in 2012 and the UK in 2015. Such services have been increasingly common in Sweden since 2004 but Europe is, sadly, way behind Japan and Hong Kong which have had 1 Gbit/s services for a couple of yeas already. Our second objective was to see if existing true broadband customers generate more traffic than conventional broadband users. If they do then it shows that the high speed of true broadband is indeed useful to the customer. The data shows that true broadband homes do generate significantly more traffic than similar homes on ADSL. Typically this is 3x the traffic per home. As higher speed access and high definition video becomes more common we expect traffic per home will grow and probably faster from fibred homes. We expect this to grow driven by larger screens, HDTV even higher resolution video and in time new applications. The message is that consumers want (and despite todays rudimentary technology already use) the greater capacity of true broadband. In future we believe they will demand even more. True broadband networks enable entirely new patterns of video consumption as already evidenced by the Swedish experience (we believe that it is no coincidence that major filesharing resources and privacy proponents are in Sweden where true broadband first reached critical mass in Europe). The growth of FTTH seems likely to generate considerable change in the audio-visual sector over the next seven years.
Page 2

Contents

Page 4 Page 8 Page 18 Page 22 Page 30

Objectives and Methodology Testing Nielsens Law Is Demand per Home Already Growing? Comparison of FTTH v ADSL Broadband Traffic Patterns Conclusions and Future Growth Drivers

Page 3

Measure actual traffic from true broadband homes. Compare to ADSL.

OBJECTIVES AND METHODOLOGY

Page 4

We Had Two Objectives


The Fibre to the Home Council Europe is a trade body concerned with promotion of investment in fibre to the home (FTTH) networks. Building such a network is very capital intensive and the question of why consumers would want FTTH is a pressing issue both for the telecom industry and also increasingly for Governments that want to ensure their own population and businesses have an up to date modern telecom infrastructure to enable success in the new digital age. Our first objective was to test Nielsens Law which was postulated in 1998. Nielsens Law of bandwidth is similar to that of Moores Law for the number of transistors on a silicon chip. It states that the bandwidth available to a high end user grows 50% per annum on average and that the mass market lags the high end user by 2-3 years. Our second objective was to examine current world observed usage behaviour by those fortunate enough to have true broadband (100 Mbit/s or more). Do looking at what such customers are doing today we wanted to see what evidence there already for the continuation of Nielsens Law (or something like it) into the fibred future? We wanted to so this because when discussing FTTH business cases and investments, two basic questions about bandwidth usually arise: Who needs all that bandwidth? What will they use it for? Note: Ventura Team LLP consults regularly in FTTH business strategy, project financing and interim management. We are a consulting and angel investment Partnership of several senior industry executives (from fixed and mobile backgrounds) and investments in various operator and service provider companies including in FTTH. We are best contacted via info@venturateam.com

Page 5

We Chose Sweden as the Focus


The Swedish FTTx market is the most advanced in Europe and the third most fibred nation in the OECD after Korea and Japan. Some years ago the Swedish Government pump primed roughly $4 billion of open fibre infrastructure investment which was amplified by Municipalities and municipally owned utilities. Around 2/3 of primary residences in Sweden are apartments which are more economic for fibre deployment than standalone houses.

OECD Broadband subscribers per 100 inhabitants, by technology, June 2007 35


Other Fibre/LAN Cable DSL

30 25 20 OECD average 15 10 5 0

Source: OECD

Page 6

Methodologies
Nielsens Law To examine the applicability of Nielsens Law we gathered historical information about the launch of new broadband services for high end consumers. To do this we studied a variety of sources (press reports, old investor presentations) and interviewed a number of senior industry contacts. In each case we determined through our research the first the year of the first reasonably wide commercial availability in at least the major city (ideally several cities) of a new broadband speed consumers service. We then plotted these results in log space using the same base data for dial-up speeds that Nielsen himself used (by its nature the telephone network and the modems of that era had very similar performance regardless of country)

Traffic on fibre networks This was the most difficult part of the study. We managed to obtain traffic information at various levels of detail from several operators and discussed patterns of usage with a larger group by means of telephone survey. We also examined what records exist in the public domain for traffic flows and patterns in the countries concerned and used interviews or published sources to determine the size and type of broadband customer bases for the different operators. By focussing on Sweden we were able to assemble a reasonably consistent dataset at the level of detail needed for this study from a base of 100,000 true broadband homes spread across various operating companies and cities. This is a large enough sample to be statistically significant when compared to ADSL in the same national market. To varying degrees through the interviews and desk research we were able to validate the general conclusions as being valid for the other countries.
Page 7

Does This 1998 `Law Still Work?

TESTING NIELSENS LAW

Page 8

Nielsens Law for Bandwidth was Similar to Moores Law for Silicon
In 1998 the web was fast growing but very new still and Jakob Nielsen was concerned with usability of web pages. Web designers tended to get carried away with the new medium and made pages too cluttered for the slow connections of the time. Nielsens point was that over dial-up it would take several seconds to load each page which was frustrating. Looking at his own Internet access speed (see top right) he postulated a Law and suggested that it would not be until 2003 that people could enjoy complex pages. His Law had two parts: A high-end user's connection speed grows by 50% per year (this is 7.5x after 5 years) The mass market lags the high-end by 2-3 years His Law implied a slower rate of progress in bandwidth than the capability of computing a trend which has in fact been realised as reported by Level 3 for the USA (see bottom right). CAGR Nielsen Moore Bandwidth Computing 50% 60% Compound over 10 years 57x 100x
Page 9
Source: the original curve from Nielsens page www.useit.com/alertbox/980405.html

Source: Level 3 Investor Conference 2007 based on various sources

Nielsens Law Works Perfectly for the UK


Our analysis of the UKs broadband history provides the speed curve shown on the right. Note that the vertical scale is logarithmic so it is nonlinear. The higher up the vertical scale the greater the difference each line represents a 10 fold increase in the speed available to a high end user. The trend rate of progress is almost exactly the 50% per annum predicted by Nielsen (in fact it is 49.87%). The curve predicts that 100 Mbit/s will be available in 2015 and 1 Gbit/s in 2020. Given the planned investment in VDSL and cable TV upgrade and the availability of LANs in many apartment blocks this seems credible bearing in mind that Nielsen refers to high end Internet users who would tend to live clustered in certain high income urban neighbourhoods with good telecom infrastructure. Note: Some University residences and even multi-tenant office blocks are already operating at 100 Mbit/s but those are small private systems and not relevant for this analysis.

Page 10

The Rate of Progress was, Unsurprisingly, Variable


This chart shows the annual rate of change in our data set for the UK. We do not have results for every year as the industry tends to progress by step change (so in some years the highest speed available was the same as hat for the year before). This is shown on the chart right where: The wavy line shows the compound annual growth rate between each observation in our data set. In other words if speeds trebled after 2 years then the CAGR would be 3^(1/2 years)=73% each year The straight line shows the trend of the data (which for the UK conforms to Nielsens predicted 50%) which is, of course, the same each year so the line on the chart is flat.

Page 11

Sweden Beat Nielsens Law


We used the same dial-up era data but plotted the curve for Swedens early deployment of Ethernet networks. The first 10 Mbit/s service became available in limited areas in 1998 and this is the date we have used. There is no hard and fast definition of geography in Nielsens Law but by 1998 we know that handful of ISPs and cable operators were offering 10Mbit/s service and that by 1999 the trend was becoming established in many areas. Similarly by 2004 100 Mbit/s was available in several areas of the country. The curve implies a high-end 1 Gbit/s service will be marketed in Sweden this year or next year. Given the simplicity of the Gigabit interface, the existing extensive residential fibre and first deployment of PONs and most importantly - growing competition in the fibre ISP space, then this seems entirely plausible.

Page 12

Sweden is 10 Years Ahead of the UK


As a way of deriving a simple message from the complicated empirical data, we plotted the trend lines for Sweden and the UK on the same chart. Looking at the positions for of the two countries for 100 Mbit/s deployment and for 1 Gbit/s we can see that Sweden leads the UK by about 10 years (or perhaps more). This is not a scientific conclusion and depends on the UK performing as poorly as predicted by what is after all a rough guide. No doubt protagonists will raise any number of detailed objections but it is clearly the case that Sweden is ahead and Nielsens Law gives us a useful, albeit simplistic, way of estimating by how much.

Page 13

France is Just Ahead of Nielsens Law and Beats the UK

France has made steady progress and beats the Law by a modest margin. Following a flurry of FTTH investment announcements last year we have assumed 2008 is the year that enough 100 Mbit/s services become available to be meaningful. Even if delayed to 2009, France has been doing better than the UK and seems set to continue to do so.

Page 14

Spain Almost Matches Nielsens Law

Our research indicates that Spain is very close to Nielsens Law. 46% CAGR we estimate compared to the predicted 50% is very close and the margin of error is probably greater than this difference. We may have missed some recent announcements. We are also aware of smaller independent networks and it may be underestimating their services.

Page 15

Poland is Doing Well but Illustrates a Simplification in Nielsens Law

Poland has extensive cable TV networks and also ~700,000 homes on local Ethernet networks though most of these are not yet connected to the Internet with fibre. There has been a great deal of activity in Warsaw and major cities to increase speeds (particularly by cable TV operators) and for new upmarket apartment developments to be provided with high speed services as part of the construction process. On this basis Poland beats Nielsens Law and does almost as well as France. However the disparity between services in the cities and the Polish countryside is huge and we believe much greater than in France, the UK etc. The reader should recall that Nielsens Law applies to high end users who, almost by definition, are in the locales with the best infrastructure. It says nothing about the equality of service throughout a country.
Page 16

An Aside An Organisation Specific Example from the USA


The University of Cedarville publishes a long run history of its connectivity on its website On discovering this we decided to test Nielsens Law at the level of a an organisation. Clearly the Law would be unreliable here because as well as variability in the telecom industrys ability to supply locally (more variable than nationally) academic budgets and many other factors come into play which fall away from a national level analysis. Despite this the results are of the right magnitude to fit Nielsens Law which, after all, was intended as a rough guide rather than a precise prediction. The speed of connectivity at Cedarville grew 69% (trend line 61%) between autumn 1993 and autumn 2007 (annual data).

Page 17

Have improvements in ordinary broadband speed had an impact on usage?

IS DEMAND PER HOME ALREADY GROWING?

Page 18

Has Bandwidth Usage by Individuals Grown?


Although Nielsens Law indicates that broadband speeds have risen and will rise further, this does not necessarily mean that actual usage in terms of the traffic volume from each home has grown at the same rate (or indeed grown at all). We have not been able to obtain detailed long run time series for a suitable set of ISPs to answer this question but we have found two international anecdotal examples (see right) and also have examined overall international bandwidth trends. International bandwidth is a separate market within the telecom industry and its growth has been documented over many years by Telegeography. Telegeographys various publications show roughly a 50% per annum rise in international bandwidth for Europe (see top right) and in 2005 they also stated that While the recent growth rates of Internet traffic are substantial, they do not approach the `doubling every 100 days calculation that was so often cited in the 1990s. TeleGeography estimates that international Internet traffic will double approximately every two years. Various studies suggest that usage has grown as the two examples from New Zealand and the USA shown right suggest. The question is that although total traffic has risen so has the number of broadband homes in Europe so has usage per home actually risen after broadband first arrived?

Page 19

The Exabyte Era, report by Cisco Systems, Jan 2008

We Estimate Traffic per Broadband Home Grew at ~20% CAGR 2002-07


1) We obtained the numbers of both dialup and broadband connections in Europe (red and blue columns respectively) 2) We then estimated traffic from a variety of sources and developed a total volume index (the green line) showing the relative amount of IP traffic from and between European countries relative to the base level in 2002. This was cross-checked against other sources (see next slide). 3) By adjusting for the mix of dial-up and broadband we estimated an indicative change in usage per broadband connection.

We believe average usage per broadband home grew almost 20% per annum over this period (about 250% in total) generated by a combination of higher line speeds and more visual web pages and other bandwidth intensive types of use.
Our analysis assumed that 1) the level of dial-up traffic per user did not materially change after 2002 on average 2) business usage stayed constant relative to consumer usage and 2) international traffic reasonably reflects growth in both total domestic and international traffic. We believe that these are reasonable assumptions in this context.
Page 20

Note: Our Underlying Estimates Seem Consistent with Observed Traffic Growth in Europe
The best sample of whole Europe traffic growth comes from www.telegeography.com Their figures are shown left. If we assume that this represents general growth reasonably well then we can cross-check and guide our bottom-up estimate to match the pattern of growth (bottom left). The two match very well. As a further cross check we looked at the main UK internet peering exchange LINX as representative of a middling speed ISP market. The total traffic growth over the period for LINX was 8x (green shaded area) remarkably close to our estimated figure for Europe as a whole of 8.2x.
Source Telegeography as presented at PTC

Page 21

Source: www.linx.net captured February 2008

More much more!

COMPARISON OF FTTH V ADSL BROADBAND TRAFFIC PATTERNS

Page 22

We Compared True Broadband to ADSL in Four European Countries


Our main research focus was Sweden because there we could examine data from around 100,000 fibred homes a far larger and representative sample than is possible at the moment anywhere else in Europe. We cross-checked our findings with fibre speed operators in three other countries. There were some differences compared to Sweden but all true broadband operators seem to experience significantly higher traffic per home compared to their local ADSL counterparts. The detail and specificity of the traffic data that we were able to obtain varied considerably by operator but the main comparative result was made at the aggregate level and we believe it is reliable and as accurate as any such estimate can be. Furthermore, the findings were confirmed and supported by the experts and managers that we interviewed across a wider set of operators in other countries.

Page 23

Fibred Homes Generate 3x the Traffic per Home of ADSL Inbound & Outbound Traffic 24 Hour Mid-Week Average
Inbound Traffic Mbits/sec/Sub - 24 Hr Average

The columns show the average throughput (as Mbit/s per customer) measured for 100k homes across 4 pure fibre operators labelled A-D. Although they would have some small businesses, most of their customers are homes. The 5th column shows the average for ADSL customers in Sweden. The red line shows the fibre average. Inbound and Outbound traffic for Pure Fibre OpCos is 3.2 to 4.4 times higher respectively compared to ADSL average suggesting both higher streaming and P2P file sharing. Anecdotal experience from some Fibre OpCos is that they are net exporters of packets driven by
Mbits/sec/Sub

0.250

0.3

0.200 Mbits/sec/Sub

3.2 Times-X ADSL Av

0.2

0.150

0.2

Fibre Average
0.100 0.1

0.050

0.1

0.000 A B C D ADSL Av.

Outbound Traffic Mbits/sec/Sub (Av) - 2200 - 00 Hrs


0.250 0.3

0.200

3.4 Times-X ADSL Av

0.2

filesharing, but our sample suggests a similar Inbound / Outbound Ratio for both Fibre & ADSL. Over the 24-hour period true broadband usage is over 3time more which is significant at this stage of market evolution given that many of the mass market applications have not been boosted to realise the advantage of Fibre.

0.150

0.2

0.100

0.1 Fibre Average

0.050

0.1

0.000 A B C D ADSL Av.

ADSL average was devised from operators with predominantly ADSL, with limited / No Fibre. Page 24

The Pattern is Slightly Higher in the Peak Hour Inbound & Outbound Traffic 22.00-00 hrs Average (Mid-Week)
Inbound Traffic Mbits/sec/Sub (Av) - 2200 - 00 Hrs

Late evening is peak time for broadband use in most European countries. We looked at the ratios for the 10pm to midnight period to see if there was any difference in peak time behaviours. Inbound and Outbound traffic for Pure Fibre OpCos are both approximately 3.8 times greater than ADSL. Outbound traffic for a Pure Fibre OpCo during the Peak-Period is 3.85 time greater than the ADSL Average, which is slightly greater than the 24-hour average (3.4 times). Inbound/Outbound Traffic Ratio for Pure Fibre OpCos is
Mbits/sec/Sub
Mbits/sec/Sub

0.250

3.84 Times-X ADSL Av

0.3

0.200

Fibre Average
0.2

0.2

0.150

0.100

0.1

0.050

0.1

0.000 A B C D ADSL Av.

Outbound Traffic Mbits/sec/Sub (Av) - 2200 - 00 Hrs


0.250 0.3

0.200

3.85 Times-X ADSL Av

0.2

essentially the same as for ADSL suggesting that end-users are using the same applications but on a larger scale. However even in Sweden it is early days for consumer true broadband - we would expect this to increase significantly as fibre become more main-stream as applications are developed with Fibre in mind.

0.150

0.2

0.100

0.1 Fibre Average

0.050

0.1

0.000 A B C D ADSL Av.

ADSL average was devised from operators with predominantly ADSL, with limited / No Fibre. Page 25

The Differences Are Less Marked but Still Significant in the Early Morning Trough Inbound & Outbound Traffic 0400-0600 hrs Average (Mid-Week)
Inbound Traffic Mbits/sec/Sub (Av) - 0400 - 0600 Hrs

Mbits/sec/Sub

4am to 6am is the trough in usage in Sweden. Most people are in bed and many backups or downloads have finished so traffic at this time if night represents the background load of large fileshares and automatically driven traffic perhaps to the USA which is in its evening peak. Inbound and Outbound traffic for Pure Fibre OpCos is 2.98 to 3.04 times higher respectively compared to ADSL average suggesting both higher streaming and P2P. Inbound/Outbound Traffic Ratio for Pure Fibre Operators is again essentially the same as for ADSL. Regardless of time of day, it is not clear whether P2P applications will significantly shift Inbound/Outbound Ratio as this depends upon the distribution of active peers and leeches in the filesharing network. Bit Torrent style protocols are evolving to better support streaming which will potentially offset any increases in Outbound Traffic due to P2P applications.
ADSL average was devised from operators with predominantly ADSL, with limited / No Fibre. Page 26

0.250

0.3

0.200

2.98 Times-X ADSL Av

0.2

0.150

0.2

0.100

0.1

0.050

Fibre Average
0.1

0.000 A B C D ADSL Av.

Outbound Traffic Mbits/sec/Sub (Av) - 0400 - 0600 Hrs


0.250 0.3

0.200 Mbits/sec/Sub

0.2

0.150

3.04 Times-X ADSL Av

0.2

0.100

0.1

0.050

0.1

Fibre Average
0.000 A B C D ADSL Av. -

Why Might Fibred Homes Generate More Traffic?


Example of Consumer Fibre Network, Morning Weekday Traditional ADSL Traffic Pattern
The traffic measurement below is taken from mid-sized ISP in a market which has a reasonable and still growing penetration of broadband. Figures show traffic in both directions for a typical day in December 2007. Browsing (inc flash from YouTube etc and http download) dominate the traffic. Off peak, a lot of traffic seems to be machine driven by peer to peer filesharing as home media servers sit there gathering the evenings entertainment. Also filesharing clients worldwide on ADSL automatically seek out the higher speed uploads available from those with fibre (so some fibre networks actually export traffic overall which is unusual for a consumer broadband network).

Note: In both cases entertainment is THE driver of demand.


The differences lie in the detail of service mix and content distribution.
Page 27

While Clearly Important, We Believe Filesharing is Not the Only Factor


TCP/UDP Prot ocol dns ssh http ftp mail voip directconnect bittorrent kazaa edonkey gnutella msnmessenger Other TCP/UD Pbased Protocol s Data 7.0 GB 4.1 MB 18.1 GB 570.6 MB 48.5 MB 5.0 MB 16.8 MB 5.7 GB 88.8 KB 12.2 MB 99.2 KB 1.8 MB 129.6 GB Flows 55,048 573 371,472 5,698 15,080 9,675 968 49,542 56 850 113 597 4,957,699 Accumulated Pe rcentage / Historical Protocol Vie w 4.4% 0% 11.2% 0% 0% 0% 0% 3.5% 0% 0% 0% 0% 80.5%

This chart illustrates the problem of measurement. As p2p traffic is designed to work around firewalls and is increasingly encrypted etc it can be difficult or impossible to identify without expensive equipment. This data is from Sweden and the interviewee told us that the Other category shown left will be almost exclusively p2p traffic of various types. However this may however be a peculiarly Swedish result. In other countries respondents believed that video streaming was an increasingly important driver of traffic and that filesharing was in some cases no more than 50%. They did, however, confirm 2x-5x greater traffic levels compared to ADSL. We briefly discuss future trends in and drivers of traffic in the next section.

Accumulated Vie w

Historical Vie w Note:

What is a flow? o TCP: a flows is a TCP connection. o UDP: a flow is a packet. TCP flows are not accounted for fully (sender and recipient) remote peers.

Page 28

Example of Benefit - Low Latency for Gamers


These Differences May Seem Slight but Affect the Purchase Decisions of Opinion Formers and so Do Impact the Market Sources of Delay in Man-machine Gaming Response in Milliseconds
300 300

ADSL is 50% slower


250 250

FTTH (with fibre grade CPE)

200

200

150

150

100

100

50

50

0 fastest right clocked processor handed eye-mouse and fast screen movement refresh ADSL firewalls ADSL line delay

0 fastest right clocked processor handed eye-mouse and fast screen movement refresh ADSL firewalls ADSL line delay

Page 29

There are no speed limits on the road to excellence.

CONCLUSIONS AND FUTURE GROWTH DRIVERS

Page 30

Can Nielsens Law Continue to Indicate the Future?


Nielsens Law was only ever intended to persuade web designers to restrict the complexity of their web pages until a reasonable proportion of US homes had broadband which he expected to happen around 2003. Despite this humble origin, a decade later it is still working fairly well as a guide to the trend in broadband speeds even in Europe. Can this ~50% pa rate of demand growth continue? We believe that it will for at least another decade driven by various factors including: Increasing file size to support consumer electronics improvements People spending more time on the Internet Diffusion of large screens going beyond HDTV and the gradual evolution of the digital home Immersive gaming and perhaps new applications who predicted the huge popularity of social networking before myspace.com?

Video walls & digital home More time online

File size

Page 31

Filesharing Has Pushed Up Average Traffic per Individual


As More Video Comes Over the Web and Files Get Larger this Trend Should Continue

http://net.doit.wisc.edu/data/flow/size/

The chart above left is of historical interest because it shows the impact on flow sizes (and implicitly traffic volume per user) over a one year period at the University of Wisconsin in the 12 month period when Napster suddenly became popular. Mp3 files were typically larger than the file types consumed in a web browsing session at that time. Napster has long since faded away as a major drive of Internet traffic but today the music charts are largely determined by Internet downloads rather than sales of CDs. As broadband speeds rise the pattern of video seems set to undergo an analogous transformation, generating much more traffic as a result.
Page 32

The chart above right from Cachelogic shows the rise of p2p filesharing traffic over a decade to dominance of volumes on mass market broadband ISP networks. Our survey of Swedish fibre ISPs indicates the p2p is nowadays an even higher proportion of traffic in an advanced market. The number of customer using filesharing remains unclear but anecdotal evidence suggests it is still a technology for the young and savvy that are prepared to pirate content. If filesharing becomes more mainstream, perhaps as the amount of legal content increases and ease of use of the software improves then it clearly has the potential to generate huge traffic volume.

As published by Telegeography in conference presentation

The Internet Has Already Partly Substituted for Traditional TV


The time budget and viewing hours result shown right is for the USA but the pattern will be similar in Europe we believe. Consumer interest and the advertising spending that it drives are shifting from TV to the Internet. In response many European broadcasters are moving online in various ways and there is a quest for viable video business models. At the same time the media industry is taking the first steps to co-existing with the reality of file sharing (cf recent content deals with BitTorrent) The stage is set therefore for video consumption over the Internet, however it works in detail, to grow steadily and drive traffic volumes ever higher.

Page 33

TV Screen Sizes are Increasing 20% Per Annum Blu Ray and Ever Higher Definition Video Will Mean Larger Files
http://www.w3schools.com/browsers/browsers_display.asp

This measurement of browser screen resolutions shows a slow but steady increase in screen resolution as PC hardware improves of around 10% each year. A similar compound rate of growth applies to consumer TV image resolutions (as measured by pixels) over the last 30 years (below right) However, the increase in television screens is much faster at the moment as a result of the replacement of CRT displays by LCD and plasma. These new technologies are completely reshaping (literally) television. According to display market intelligence experts Meko, the weighted average size for a TV set in Europe will increase from around 26" at the beginning of 2006 to just over 34" at the end of 2009. This is a CAGR of 20% (by screen area) and with increasing sales of 70 inc screens and the launch of 80+ inch next year this trend seems set to continue over the medium term. Broadband will need to support larger and larger file sizes to fit high definition programming and larger, higher resolution screens

Source: Ventura Team derived from Wikipedia data

Page 34

Ciscos Detailed Bottom-Up Projections Assume Entertainment Drives Volume Growth per Home (augmented by new apps and mobile)
Ciscos Global IP Traffic Forecasts are based on very detailed bottom-up analysis of drivers and trends including assessment of historical behaviour. Their model projects growth in European traffic volume of 8x9x over the 5 year period 20062011 This is roughly consistent with our own analysis of past trends and equates to a CAGR of ~50% per annum

Source: Global IP Traffic Forecast and Methodology 2006-2011, Cisco, 2008

Page 35

Key Messages
This study has shown that: As a general guide, European broadband speeds are rising around 50% per annum and should continue to do so. This is Neilsens Law in action. When customers have faster connections they use them more. This is true for conventional broadband and even more so for fibre in fact fibre customer use broadband (in terms of traffic) about 3 times more than ADSL customers and it is quite possible that this difference may grow further over time because it is still very early days for the fibre speed Internet. It takes time for the higher speeds to spread from the aficionado to the mass market but they do and traffic keeps growing. Broadband traffic per home has been growing at 20% pa over the last few years and this will continue and the growth rate will probably accelerate significantly both as video over the Internet becomes more common and also more homes are provided with true broadband. We believe that these trends have a number of implications: Operators will need to invest and upgrade to take true broadband into the home. Eventually customers will simply require this but the most rapid progress will be made in markets with high infrastructure competitive pressure or where a regulator acts so as to simulate such pressures. Investors will need to understand the nature and potential operational / regulatory and financial structures of this 100 trillion wave of capital investment in fibre access across Europe. Governments (local and national) have two issues to consider in countries which are only on or below Nielsens curve, are they happy to trail behind competing nations and of not, what is the best pump priming or other action they can take? Within country, how great a divide between the best served areas and the worst is tolerable and what can be done to minimise it. Customers can look forward to more speed and bigger better pictures while media companies face a period of continuing change and need to respond to the impact that high speed Internet distribution and ever falling local storage costs are having on consumer behaviour and content distribution..
Page 36

Potrebbero piacerti anche