Description
Over the last decades, the nature of Internet traffic has changed from static data & text to interactive media content, effectively
transforming the Internet into a new media platform as its usage shifted to richer types of content, particularly streaming video.
The future development of the Internet as a media platform is nowadays challenged by increasing global connectivity, proliferation
of smart devices and streaming media services, which cause spectacularly higher traffic volumes, greater traffic imbalances and
changing traffic patterns. Internet traffic doubles almost every two years, and traffic patterns have changed as a result of real-time
streaming overtaking peer-to-peer as the predominant form of digital distribution. Internet access networks experience significant
in the order of 5:1 imbalances between incoming and outgoing traffic because of the media-related nature of traffic, which mainly
flows one way, from content providers to end-users.
In 2014, the Internet reached over 2.7 billion individuals and has become mission critical for most Content and Application Providers.
Indeed, minor disturbances in the quality of delivery directly impact the willingness of end-users and advertisers to pay for online
services.
Innovation and Investment in IP Interconnection
The Future of the Internet
May 2014
Content
Foreword 3
Key Messages 5
Executive Summary 6
1. The Internet is Vital and Continuously Mutating 11
1.1. The Internet is now mission critical commercially and calls for quality delivery 11
1.2. The IP Interconnection evolves and follows the evolution of the Internet 16
1.3. Introducing the actors in the IP Interconnection value chain 19
1.4. Description of IP Interconnection business models 21
2. So far, the IP Interconnection Value Chain has Adapted Well 31
2.1. Content and Application Providers and Terminating ISPs are setting the pace of
IP Interconnection innovation 31
2.2. Internet Content and Application Providers look for quality control 33
2.3. A new power emerged: the arrival of the Internet Global CAPs 35
2.4. Investment strategies for future IP Interconnection 36
2.5. Friction occurs in the IP Interconnection value chain, but is (so far) quickly resolved 38
3. Revolutionary Future Applications that Require New Delivery Features 40
3.1. The future Internet driven by Internet of Things and Internet of Humans? 40
3.2. Today’s Internet is a Best Effort and finite (yet-not-scarce) resource 43
3.3. New requirements are emerging, beyond bandwidth 48
3.4. New IP Interconnection business models are being developed 52
4. New Business Models Could Accelerate Innovation & Value Creation 55
4.1. The most advanced application landscapes could generate substantial economic value creation 55
4.2. The acceleration of the most advanced application landscapes require to guarantee a link
with the most suitable Internet platform option 56
5. Three Core Assumptions can Drive the Evolution of the Future Internet Platform 59
5.1. The future Internet Platform will, as always, organically grow from stakeholders various interests 59
5.2. Three core assumptions have the potential to shape the future Internet platform 60
5.3. Looking forward, three Internet platform options can be foreseen:
Best Effort 2.0, Quality-Guaranteed Services, Both Worlds 63
Glossary 67
3
Scope of this study: “The Internet” and “IP Interconnection”
The future of the Internet is a widely debated public policy theme all over the world. Questions are raised on
how to preserve the public “best-effort” Internet as an “open” platform for innovation and competition, and
how to combine economic- & societal value creation and sustainable returns on investment. Although so far
the history of the Internet has been an incredible success in organically developing a self-adapting complex
of network business relations, concerns are raised about how the Internet will be able to sustain an adequate
quality of experience for the end-user in the future. This ability may come under pressure by a spectacular
boom in Internet traffic volumes in the coming years, resulting in unprecedented demand for reliable,
ubiquitous Internet access and mass uptake of bandwidth-intensive services and applications. To illustrate
this point: by 2020, more than 50% of the world’s population will be online. This means an increase from 2.7
billion users in 2014 to 5.0 billion users by 2020. By 2025, “The Internet of Things” will comprise around 50
billion connected devices. By 2030, machine-to-machine (“M2M”) communication is expected to constitute
more than 50% of IP traffic.
The question therefore seems justified as to whether the Internet can cope with this evolution, and
who and what is needed for the Internet to evolve and adjust to these changing circumstances.
One part of the answer lies in capacity, quality and traffic management in the Internet access network (fixed
or mobile), or so-called “last mile”, owned or operated by an Internet Access Provider over which end-users
access the Internet. This part is the subject of “net neutrality” discussions and mainly covers the front end,
consumer-facing side of the Internet.
The other part of the answer lies in the so-called “up-stream” side of the Internet. This is where the
Internet access networks connect with (i) each other, (ii) bulk IP traffic transportation networks and undersea
cables connecting continents and (iii) content & application server parks located across the globe. This “IP
Interconnection” part of the Internet solely consists of wholesale agreements, which determine the technical
& economic conditions under which IP traffic is delivered from the originating party (for example, a content
& application provider or an ISP). This is done via several exchangeable delivery networks of multiple Internet
connectivity providers (often used in parallel) to the residential Internet access networks of terminating ISPs,
and vice versa.
IP Interconnection is, and has been, an essential building block for the quality and functionality of the Internet
as ultimately experienced by the end-user, despite the fact that the end-user is no party to IP-Interconnection
arrangements. IP Interconnection models have adapted to changes in Internet usage and traffic patterns
caused by disruptive applications or technologies (for example, predominantly digital distribution technology
moving from decentralized peer-to-peer to centralized streaming) and facilitated IP content delivery
accordingly. In many ways, the extent to which the IP Interconnection sector is able to innovate itself defines
the scope of evolution of the Internet as a platform for future applications.
Foreword
Therefore, IP Interconnection developments have an impact on overarching objectives in Internet public policy debates that,
essentially, focus on warranting end-users’ quality of experience over the public Internet, or “Best-Effort” Internet.
With this report, we want to unravel some of the complexity in IP Interconnection and identify the main drivers of change in IP
Interconnection. We also analyze the effects of investment and innovation by the IP Interconnection players on the future capability
of the public Internet, end-user quality of experience, competition and the scope for new Internet applications.
In order to achieve this, the study first analyzes the latest stage in the evolution of the Internet, and then the corresponding
trends in IP Interconnection reflecting this evolution, asking if the IP Interconnection players will be able to continue to reach
innovative interconnection business models and participate in the Internet value chain. Finally, the study analyzes how different IP
Interconnection scenarios may affect the future of the public “Best-Effort” Internet, which is linked to unlocking the potential of
innovation and new, exciting application paradigms such as the “Internet of Things” and the “Internet of Humans”.
Sincerely,
Gregory Pankert
Partner
TIME* practice
Arthur D. Little
Andrea Faggiano
Principal, Head of SASCAR**
TIME* practice
Arthur D. Little
Karim Taga
Global Practice Head
TIME* practice
Arthur D. Little
* TIME: Telecommunication, Information Technology, Media and Electronics
**SASCAR: Strategic Advisory Services for Competition and Regulation
The Future of the Internet
5
The Internet is vital, continuously evolving and has developed into a new media platform
1. The Internet has been transformed into a new media platform, as the nature of Internet traffc has changed from static data &
text fle transfer to streaming interactive media content.
2. The Internet has become mission critical for most Content & Application Providers. Minor disturbances in the quality of delivery
directly impact the willingness of end-users and advertisers to pay for online services.
3. The future development of the Internet as a media platform is impacted by increasing global connectivity, proliferation of smart
devices and streaming media services which cause spectacularly higher traffc volumes, greater imbalances in traffc fow and
changing traffc patterns.
4. IP Interconnection is an essential building block for the quality & functionality of the Internet as ultimately experienced by the
end-user, despite the fact that the end-user is no party to business-to-business IP Interconnection arrangements.
IP Interconnection, so far, adapted well to support the changing nature of the Internet, and remains dynamic and
competitive
5. The IP Interconnection value chain converges, but remains dynamic and competitive. Proliferation of Content Delivery Networks
and Internet Exchanges, commoditization of IP transit and CDN prices challenge existing interconnection models and enable
new ones.
6. From the early days of “IP transit” and “Peering”, a genuine mix of viable application/content delivery strategies is accessible to
all players seeking connectivity.
7. Content & Application Providers and ISPs are setting the pace and determining the nature of IP Interconnection innovation by
vertically integrating and interconnecting directly, which disintermediates pure Internet connectivity providers to some extent.
8. Changes in the IP Interconnection ecosystem lead to tension between IP Interconnection players. However, disputes concern
less than 1% of all IP Interconnection agreements and are solved without regulatory intervention in more than half of these
cases.
9. End-users have not been substantially or structurally affected by IP Interconnection disputes.
Future applications require IP Interconnection models to evolve towards providing higher-quality assurances,
which will impact the current “best-effort” Internet
10. Innovation in IP Interconnection is needed to support further development of the Internet and accelerate take-up of next-
generation applications (Internet of Things, Internet of Humans) that require IP Interconnection Quality of Service (latency, jitter,
packet loss) extended with new parameters (e.g. security, data protection).
11. Variants of Paid Peering, Deep Caching, Assured Delivery or Secure M2M are among the innovative IP Interconnection business
models that could lay the foundation for an advanced Internet platform, based on assured end-to-end Quality of Service Internet
Platform – complementary to “Best Effort”.
12. “Best-effort” Internet is and will no doubt continue to be essential in the future, and there is early evidence to indicate that it can
continue to improve and coexist with complementary end-to-end Quality of Service platforms if properly monitored.
13. Private investment in IP Interconnection has led to structurally improved conditions for the future development of the public
Internet. Content comes closer to end-users (by direct interconnection and local content caching), Internet performance is
improved by adoption of new application technologies (e.g. “adaptive streaming”) and IP network resources are abundant (e.g.
higher capacity in the “last mile”).
Key Messages
The Future of the Internet
6
The Internet is vital, continuously evolving and has developed into a new media platform
Over the last decades, the nature of Internet traffic has changed from static data & text to interactive media content, effectively
transforming the Internet into a new media platform as its usage shifted to richer types of content, particularly streaming video.
The future development of the Internet as a media platform is nowadays challenged by increasing global connectivity, proliferation
of smart devices and streaming media services, which cause spectacularly higher traffic volumes, greater traffic imbalances and
changing traffic patterns. Internet traffic doubles almost every two years, and traffic patterns have changed as a result of real-time
streaming overtaking peer-to-peer as the predominant form of digital distribution. Internet access networks experience significant
in the order of 5:1 imbalances between incoming and outgoing traffic because of the media-related nature of traffic, which mainly
flows one way, from content providers to end-users.
In 2014, the Internet reached over 2.7 billion individuals and has become mission critical for most Content and Application Providers.
Indeed, minor disturbances in the quality of delivery directly impact the willingness of end-users and advertisers to pay for online
services.
IP Interconnection, so far, adapted well to support the changing nature of the Internet, and remains dynamic
and competitive
IP Interconnection is an essential building block for the quality & functionality of the Internet as ultimately experienced by the end-
user, despite the fact that the end-user is no party to business-to-business IP-Interconnection arrangements. IP-Interconnection
models adapt to changes in Internet traffic patterns caused by disruptive applications, or technologies, and facilitate IP content
delivery accordingly. In many ways, the extent to which the IP Interconnection sector is able to innovate itself defines the scope of
evolution of the Internet as a platform for future applications.
Executive Summary
Figure A - The changing nature of the Internet and the new capacity/quality requirements
* downstream traffic
COMMUNICATION: services provided by application (Skype, WhatsApp, iMessage, FaceTime,etc.); DATA: file sharing (Bit Torrent, eDonkey, etc),
web browsing, social networking, email, etc.; MEDIA: streamed or buffered audio and video (Netflix, non-linear TV services;
1. 2009-2012 CAGR; 2. Interviews. Source: ITU, Sandvine; Arthur D. Little analysis
More
bandwidth
More quality
(e.g. reduced delay)
Media
Communication
Data
Peer-to-peer
download
5:1
In/out traffic
unbalance
2
+39%
yearly traffic
growth
1
na
na
21%
10%
33%
11%
33%
14%
Data
Media
(streamed or
buffered one-
way audio and
video)
Communi-
cation
% of fixed traffic at busy hour (North America illustration)
*
*
32%
43%
53%
62%
68%
54%
45%
36%
2011
2% 2%
2010
3%
2009 2012
The Future of the Internet
7
The Internet and the underlying IP Interconnection ecosystem demonstrated an organic ability to evolve and adapt. Alternative
business models (such as peering and Content Delivery Networks) challenged existing ones and improved the overall efficiency of IP
Interconnection, leading to a cost reduction of around 30% per annum over the last decade.
From the early days of “IP transit” and “Peering”, a genuine mix of IP Interconnection models is currently available to both ISPs and
Content & Application Providers (CAPs) seeking connectivity. This is the result of three major developments:
1. Decentralization of the Internet: the emergence of national and regional Internet Exchanges facilitates private peering
arrangements by increasing number of ISPs’ access networks edges in one central location.
2. Commoditization of IP Interconnect prices (falling IP transit, CDN or router costs) led to substitutability of IP Interconnection
products and countervailing powers in the IP Interconnection value chain.
3. Proliferation of Content Delivery Networks: Content & Application Providers leverage the increased value of their Internet
content by building proprietary caching server parks or, alternatively, using independent, commercial CDN services that are
located close to the ISPs’ access networks.
Still, the majority of Internet traffic is progressively being concentrated to a limited number of large Content & Application Providers
and a few wholesale carriers. In 2013, 35 networks carried 50% of all Internet traffic in North America, down from 150 networks in
2009. The concentration of IP traffic is a major evolution in the IP Interconnection value chain, and has the potential to influence the
negotiating power among connectivity stakeholders and affect the current equilibrium in the Internet ecosystem.
In the last years, the largest Content & Application Providers and Internet Service Providers have been setting the pace and
determining the nature of IP Interconnection innovation through vertical integration. Content & Application Providers seek end-
user proximity and are increasingly investing in proprietary Content Delivery Networks or relying on third-party CDNs. ISPs invest
in network-based content delivery platforms (“deep caching”) for internal purposes and as a service to third-party Content &
Application Providers.
As a result, Content & Application Providers and Internet Service Providers increasingly interconnect directly, disintermediating pure
Internet connectivity providers to some extent. Improving control on the quality of delivery over the Internet is the main motivation.
This is true not only for Internet-based CAPs, but also increasingly for the video-streaming strategies of traditional broadcasters
(e.g. BBC iPlayer’s average daily unique users grew 33% year on year since 2009). The equilibrium in the IP Interconnection value
chain has subsequently changed, and traditional IP Interconnection players have adapted to maintain their competitiveness. Internet
Figure B – Traffic concentration trends (2007, 2009, 2013)
* 2013 figures refer to North America Internet traffic
Source: Deepfield; Arthur D. Little analysis
Akamai
40%
Google
20%
0
15
30
45
60
75
0 40 80 120 160 200 240 280 320 360 400
2013*
2009
2007
% of North America traffic
# of networks
50% of traffic from
1000s of networks
50% of traffic from
150 networks
50% of traffic from
35 networks
% of North America traffic
The Future of the Internet
8
connectivity providers such as IP Transit providers, independent CDN providers and Internet Exchanges are under pressure to
innovate and diversify their service offerings (e.g. offering “partial transit”, commercial open CDNs or web security) or attract a critical
mass of traffic through consolidation (e.g. international carrier Level 3 acquiring its competitor, Global Crossing).
Changes in the IP Interconnection ecosystem meant that tensions between IP Interconnection players intensified. However,
disputes concern less than 1% of all IP Interconnection agreements, and are solved without regulatory intervention in more than
half of these cases. There are a number of reasons for this:
n
IP Interconnection (including upgrade) costs account for just a marginal share, i.e. less than 1% of the overall connectivity costs.
n
Countervailing powers emerged by changing the IP Interconnection economics that keep the value chain in balance:
Figure C – Trends over the IP Interconnection value chain
Source: Arthur D. Little analysis
Drivers
Strategic
moves
Large CAPs seek for user proximity
END-
USERs
? Grow scale
? Defend profitability
and investment
payback
? Secure quality of
service for own
applications
? Search for
economies of scale
? Reach/keep critical
mass
? Attract new
members
? Diversify revenues
? Grow scale
? Defend profitability
? Look for new
revenue streams
and monetize
eyeballs
Partnering with
access providers
Partnering with
access providers
own
infrastructure
own
platforms
own
infrastructure
own
platforms
Core Business
own
platforms
Core Business
own
platforms
IP Transit
providers
Content Distribution
providers (CDN)
Terminating ISPs
Interconnection
Exchanges (IX)
Content &
Application
providers
Explosion of Internet
Exchange and Peering
Significant drop in
router prices
Explosion of Content
Delivery Networks
Commoditization of IP Transit
and CDN prices
Router cost per Gbps ($)
CDN video prices ($/GB)
IP Transit prices ($/Mbps)
0
1
2
3
4
5
6
7
8
9
10
11
12 1.20
1.05
0.90
0.75
0.60
0.45
0.30
0.15
0.00
2013E
0.05
1.6
2012 2011 2010 2009 2008
0.30
12.0
-33%
-30%
30
3
122
2010 2008 2006
25
2012
155
Global # of CDN operators
30%
Largest CAPs’ CDN
Commercial CDN
Network Operators’ CDN
Global # of IXs by region
2
0
1
1
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
2
0
0
7
2
0
0
8
2
0
0
9
2
0
1
0
Latin
America
North
America
Europe Asia
0
47.000
1.000
2007 2012 2003 1999
50
100
150
250
300
350
400
Figure D – Economic drivers of IP Interconnect evolution and countervailing powers
Source: ITU, Informa, Packet Clearing House, Dr. Peering, Cisco, streamingmedia.com, Web sites, Arthur D. Little analysis
-24%
The Future of the Internet
9
a) IP Transit and Peering have become substitutes in terms of cost
b) Falling IP Transit and CDN prices make high-quality transport and CDN strategies accessible to smaller CAPs
c) Falling IP Transit prices balance against paid-peering cost pressures
d) Strong retail competition prevents market foreclosure by ISPs
End-users have not been substantially or structurally affected by IP Interconnection disputes. The commercial interest of parties
prevailed, and mutually acceptable solutions were found. Associated Interconnection costs have not proven to be prohibitive to core
business models so far.
Future applications require IP Interconnection models to evolve towards providing higher-quality assurances,
which will impact the current “best-effort” Internet
In addition to continuous improvement of connectivity between Content & Application Providers and access networks, innovation in
IP Interconnection can support further development of the Internet and accelerate the take-up of next-generation applications that
require uncompromised quality. In particular, the Internet of Things and the Internet of Humans application landscapes can unlock
an economic value potential in the range of trillions of euros by 2020. However, advanced Internet platforms, i.e. beyond Best-Effort,
may be required for next-generation applications, which could bring an Internet of Things and an Internet of Humans to life.
IP Interconnection Quality of Service needs to be extended to new parameters (e.g. latency, jitter, packet loss, security, and
data protection). As the Internet evolves from nice-to-have services to mission-critical services, next-generation applications for
sectors such as the Financial Services industry, the Electronic Payment sector, high-security Governmental Bodies (police, military,
emergency services, etc.) will generate a demand for new IP Interconnection requirements going well beyond additional throughput
capacity. It will expand to delivery features relevant for streaming video, such latency reduction, availability, jitter control and packet-
loss limitation. Security and data protection deserve special attention as they play a critical role in the safe use of next-generation
applications, especially in scenarios foreseeing the Internet of Things leading to M2M applications such as connected cars with
remote-start features.
Variants of Paid Peering, Deep Caching, Assured Delivery and Secure M2M are among the innovative IP Interconnection business
models that could lay the foundation for an advanced Internet platform, based on an assured end-to-end Quality of Service Internet
platform – complementary to Best-Effort.
Figure E – Innovation in IP Interconnect business models
Source: Arthur D. Little analysis
Best-Effort High-Quality
Openness
Offered service
Assured quality
Reporting services
Interconnection
point
Application
risk sharing
None
Selected networks
Secure
networks
Transfer +
close caching
None
None
Co-design
risk sharing
Commercial launch
risk sharing
Peering IP Transit Deep Caching
new
Paid Peering Assured Delivery
new
Secure M2M
new
IP Interconnection
dimensions
Entry IP port
(upstream )
All networks
Transfer
extra threshold
Balanced
transfer
Caching +
Transfer
IP port in the Local Exchange
(downstream, closer to users)
Guarantee
on port
availability
Guarantee on
access transfer
(delay, jitter)
Guarantee on
transfer (by traffic
classes)
Reporting QoS between
IP port and access
gateway
Reporting QoS
between IP port
and user device
Reporting on
QoS at IP port
Transfer +
hosting locally
The Future of the Internet
10
Still in the advent of new IP Interconnection business models, the Best-Effort Internet is, and will no doubt, continue to be essential
in the future, and there is early evidence to indicate that it can continue to improve and co-exist with complementary end-to-end
quality of service platforms if properly monitored. Best-Effort has long co-existed with business-to-business IP managed services,
as well as with ISPs’ managed IPTV platforms, and its average and peak connection speeds still increased by respectively 12% and
23% since 2007, increasing to 21% and 26% since 2011.
Private investments in IP Interconnection resulted in a number of trends with structurally improved conditions for the future
development of the public Internet:
a) Content comes closer to end-users: Direct interconnection with fewer Internet connectivity providers allows for better
structural conditions on Quality of Service - i.e. lower latency, lower risk of packet loss and jitter.
b) New application technologies improve performance: This includes codecs, adaptive streaming and content distribution
algorithms.
c) Abundance of IP network resources: Capacity in the “last mile” became larger. IP transit and Content Delivery services
were being commoditized, and new quality delivery opportunities were created, with network-based “deep-caching”
technologies.
d) High dynamism in the Internet ecosystem: The high value at stake for all stakeholders allows new disputes to be resolved
quickly and create new business relationships.
The public Internet will stand to benefit mostly from private investments in IP Interconnect architecture aimed at shortening the
distance that Internet traffic needs to travel before it reaches the “last-mile” Internet access networks. This is accomplished by
storing popular content/applications in local servers that form part of proprietary or commercial CDNs, or in network-based “deep-
caching” servers. Shorter travel distances for IP content with fewer intermediaries implies less chances for “bumps” in the road.
It also increases the prospects for more manageable end-to-end controls, leading to an overall higher quality of experience for the
end-user.
Figure F – IP Interconnection structurally contributing to Internet’s improvement
Source: Akamai, Arthur D. Little analysis
IP Interconnection trends:
better structural conditions
Best-Effort Internet is continuously improving
? Content comes closer to end-users
(direct interconnection with fewer intermediaries, allowing
for better structural conditions on QoS: lower latency, lower
risk of packet loss and jitter)
? New application technologies improve performance
(codecs, adaptive streaming, content distribution
algorithms)
? Abundance of IP network resources
(larger pipes in the last mile, IP Transit and Content Delivery
services being commoditized, new opportunities with deep
caching technologies)
? High dynamism in the ecosystem
(the high value at stake for all players allows the new
disputes to be resolved rapidly and new business
relationships to be created)
0
1
2
3
4
5
6
7
8
Q2’08 Q4’07
+21% p.a.
+12% p.a.
Q4’13 Q2’13 Q4’12 Q2’12 Q4’11 Q2’11 Q4’10 Q2’10 Q4’09 Q2’09 Q4’08
AMERICAS EMEA GLOBAL
Average connection speed
The Future of the Internet
11
1.1. The Internet is now mission critical
commercially, and calls for quality delivery
The Internet rapidly and dynamically evolved from an
experimental network to a mass-market interaction
platform
The story of the Internet began in 1957, when the USA
responded to the USSR, launching Sputnik into space with the
creation of the Advanced Research Projects Agency (ARPA). The
Agency’s mission was to become the leading force in science
and new technologies. Around 1994 the Internet was used for
the first time as an open commercial platform, enabling the
launch of the first Internet ordering system (Pizza Hut) and the
first Internet Bank (First Virtual). From that point on, all the major
Internet companies emerged: Google was launched in 1998,
MySpace in 1999, Apple’s iTunes Store in 2003, Facebook in
2004, YouTube in 2005 and Twitter in 2006. Netflix (created
in 1998) started to offer online streaming services in 2008.
Soon, their rapid growth and potential success would bring
these Internet companies to become listed companies, Google
showing the lead in 2004. Facebook (2012) and Twitter (2013)
made remarkable entries to the stock market.
From its birth in 1969 to nowadays, the Internet has evolved and
mutated in many different ways:
n
Internet usage ceased to be a US-centric phenomenon and
achieved a global reach;
n
Traffic boomed and regional poles emerged, while the nature
of Internet content progressively changed from static text
and simple data to interactive media and entertainment;
n
The way of accessing the Internet shifted from dialing in via
fixed networks to an always-on mobile experience
n
New network requirements emerged, and quality of
delivery – which, in the early stages was not that important –
became mission critical.
The popularity of the Internet grew substantially, and the
penetration of Internet users within the world’s population
strongly increased.
Over the 2005-2013 period, the number of individuals using the
Internet
1
grew yearly at 13%, from 1 billion to almost 3 billion,
mainly outside the US.
1 Source: ITU; an Internet user is someone aged 2 years old and above who
went online in the past 30 days
1. The Internet is Vital and Continuously
Mutating
Figure 1: The Internet timeline
Commercial age: 25 years, so far Research age: 30+ years
1960 1965 1980 1985 1995 2010
1957: The USA creates the
Advanced Research Projects
Agency (ARPA)
1969: First switched
network (called ARPANET)
is created, connecting four
different nodes in
California and Utah
1975
1989: ARPANET
ceases to exist
1993: the first web browser,
Mosaic (later Netscape) is
released
1990
1994:
? First Internet ordering system
(pizza Hut) is launched
? First online payment system
(First Virtual) is launched
2005: YouTube is
launched
2005
1998:
Google is
launched
1999: MySpace
is launched
2003: Apple launches
iTunes store
1970: NCP protocol is
developed as
ARPANET’s host-to-
host protocol
1973: First written
version of the TCP/IP
is developed
1976: SATNET, a satellite
program, is developed to link
the USA to Europe
1970
2005:
Facebook is
launched
1999:
Wi-Fi is
standardized
2012: Facebook
becomes a listed
company
2000
1983: Domain
Name System
is introduced
1983: TCP/IP
becomes the
standard
internet
protocol
2008: Netflix starts
to offer online
streaming services
1972: Ethernet
is established by
Metcalfe
1972: First program
devoted to email is
created
1989: CERN releases the
World Wide Web
1990: The first search engine
is created, called the Archie
Search Engine
1996: Nokia
releases the
first cell phone
with Internet
access
2006: Twitter is
launched
2013: Twitter
becomes a listed
company
2004: Google
becomes a listed
company
Source: Arthur D. Little
The Future of the Internet
12
Regional data from ITU reveals that this growth is consistent
across all world regions. Europe’s and America’s relatively low
growth during these recent years (7% and 8%, respectively) is
explained by their earlier take-up of the Internet, and both regions
together now account for only 38% of global Internet users.
However, Internet traffic origination is not equally distributed
across global regions: most of the IP traffic originates in North
America (34%), followed by Asia & Pacific with 33%, Europe
with 24%, Latin America with 8% and the Middle East & Africa
with only 2%.
CISCO indicates that IP traffic will continue to grow, but at lower
growth rates: highest growth rates are expected from 2013 to
2017 in the Middle East & Africa (35%) and Asia & Pacific (24%).
In the beginning, Internet access relied mainly on fixed-network
infrastructure and end-users experienced the Internet through
desktop computers. Over the years, a vast array of connected
devices emerged, such as smartphones and tablets, through
which end users could experience the Internet in mobility.
These events significantly increased Internet’s accessibility and
boosted its penetration in the global population.
The number of broadband (BB) subscriptions has grown over
2007-2013 at a growth rate of 28.7 %, from 614 million to 2.8
billion, but when looking closer at their composition, it is clear
how the mobile broadband development did become the main
driver of this significant growth trend, especially in developing
countries
2
.
Between 2007 and 2013, the share of mobile broadband access
increased from 43% of total broadband subscription to 75%.
As the number of Internet users and broadband subscriptions
has increased, the volumes of IP traffic carried over the Internet
2 Classifcation of developing and developed country is available athttp://www.
itu.int/ITU-D/ict/defnitions/regions/index.html, and has been made according
to UN M49
Figure 2: Individuals using the Internet
Source: ITU, Arthur D. Little analysis
44%
2013
2.7
46%
21%
17%
5%
22%
19%
5%
5%
4%
2010
2.0
43%
23%
21%
5%
5%
4%
2009
1.7
42%
25%
22%
4%
4%
4%
2008
5%
40%
26%
24%
4% 4%
3%
2007
1.3
37%
29%
25%
3% 3%
2%
2006
1.1
35%
31%
26%
3%
3%
5%
2005
1.0
34%
31%
27%
3%
3%
2%
5%
1.5
2012
2.5
45%
22%
18%
5%
5%
2011
2.3
2%
Individuals using the Internet (Bln) CAGR
Europe +7%
Asia & Pacific +18%
The Americas +8%
CIS +23%
Middle East +23%
Africa +30%
Figure 3: Global IP traffic by region
2017E 2012 2016E 2015E 2014E 2013E
120.643
14.831
101.055
83.837
10.174
55.553
43.570
2011 2010 2009 2008
68.892
28.046
20.182
Global Internet traffic (PB/month) CAGR
(’08-’13)
+57%
+70%
+36%
+33%
+49%
North America
Asia Pacific
Latin America
Middle East and Africa
Europe
Source: CISCO, Arthur D. Little analysis
The Future of the Internet
13
have grown tremendously. Cisco Visual Networking Index shows
that traffic has risen from 10.1 exabytes (1018 bytes, or 1 billion
gigabytes) per month in 2007 to 55.5 EB/month in 2013, with a
40% CAGR, and is expect to reach more than 120 EB/month in
2017 (with a growth rate of 21% from 2013 to 2017).
Fixed-Internet traffic accounts for about 70% of Global IP traffic,
and has grown at a CAGR of 37%, reaching 39 EB/month. It is
expected to reach 82 EB/month in 2017, counting for roughly
68% of the total IP traffic.
Managed IP services such as IP Virtual Private Networks (IP
VPN), historically applied in business-to-business environments,
have also grown significantly from 2007, at a CAGR of 49%.
These have reached 15 EB/month in 2013, accounting for 26%
of total IP traffic; CISCO projections in 2013 indicate that IP
managed services traffic could reach 27 EB/month by 2017,
slightly decreasing its share over total Internet traffic (23%).
Global mobile traffic has grown at notable rates (CAGR of 117%),
reaching 1.6 EB/month in 2013, but it still accounts for only 3%
of total IP traffic; however, CISCO expects it to rise to 11 EB/
month by 2017, accounting for more than 9% of total IP traffic.
This growth of IP traffic volumes – data consumption per user
grew from 12 gigabytes per user per month in 2008 to roughly
Figure 4: Internet Broadband Subscription and Global IP traffic by type
Source: ITU, Arthur D. Little analysis
411
468
527
588
638
422
615
778
1.556
696 Fixed broadband
Mobile broadband
2013E
2.792
2.096
2012
2.194
2011
1.743
1.155
2010
1.305
2009
1.083
2008
833
Broadband subscriptions (mln)
+12%
+41%
Fixed Internet
2013E
Managed IP
Mobile data
55.552
2012
43.570
2011
28.035
2010
20.181
2009
14.832
2008
10.174
Global Internet traffic (PB/month) CAGR
(’08-’13)
+117%
+49%
+37%
CAGR
(’08-’13)
Figure 5: The changing nature of the Internet and the new requirements
* downstream traffic
COMMUNICATION: services provided by application (Skype, WhatsApp, iMessage, FaceTime,etc.); DATA: file sharing (Bit Torrent, eDonkey, etc),
web browsing, social networking, email, etc.; MEDIA: streamed or buffered audio and video (Netflix, non-linear TV services;
1. 2009-2012 CAGR; 2. Interviews. Source: ITU, Sandvine; Arthur D. Little analysis
More
bandwidth
More quality
(e.g. reduced delay)
Media
Communication
Data
Peer-to-peer
download
5:1
In/out traffic
unbalance
2
+39%
yearly traffic
growth
1
na
na
21%
10%
33%
11%
33%
14%
Data
Media
(streamed or
buffered one-
way audio and
video)
Communi-
cation
% of fixed traffic at busy hour (North America illustration)
*
*
32%
43%
53%
62%
68%
54%
45%
36%
2011
2% 2%
2010
3%
2009 2012
The Future of the Internet
14
20 gigabytes per user per month in 2013 – is not only driven
by the increase in the number of broadband subscribers, but is
also, and most importantly, linked to the change in nature of the
Internet traffic itself.
The nature of Internet traffc changed from static data
and text to interactive media content
The latest measures available show that the bulk of total usage
growth comes from real-time streaming devices; the Internet
has transformed itself from a data- and file-transfer platform into
a new-media platform, and its usage has shifted to richer types
of content, particularly video.
Today more than 60% of Internet traffic in the US is media
related and, seemingly, such share is expected to grow further
in the coming years
The consequences of such a shift materialized into increased
demand for a higher bit rate and delivery quality. During the
early ages of the Internet, communication interactions were
established through sequential (asynchronous) applications, and
on-time delivery was not important (e.g. emails); nowadays,
higher throughput and reduced delivery time are essential for a
good quality of experience (see figure 5 overleaf).
Furthermore, traffic kept doubling almost every two years,
and traffic patterns changed as a result of real-time streaming
Figure 7: New connectivity requirements
Source: Arthur D. Little analysis
Uncontended capacity
More quality
(e.g. reduced delay or packet losses)
P2P applications
Near-real-time
streaming
Download Streaming
One-way
live streaming
Live football
matches
Prime-time events
Two-way
live streaming
HD
Ultra HD
HD
HD
Figure 6: Daily Internet traffic by content type
Note: Storage and Back-up Services: PDBox, Netfolder, Rapidshare, etc..; Bulk Entertainment: iTunes, movie download services; P2P Filesharing: BitTorrent, eDonkey, Gnutella,
Ares, etc…; web Browsing: HTTP, WAP browsing; Real-time Entertainment: streamed or buffered audio and video, peercasting, specific streaming sites service (Netflix, Hulu,
YouTube, Spotify,..);
Source: Sandvine, Arthur D. Little analysis
? 60% of Internet traffic consists of
video (real-time entertainment +
bulk entertainment) during most of
the day
? During night hours Internet traffic
reaches its peak
Outside top 5
Storage and Backup services
Bulk entertainment
P2P file sharing
Web browsing
Real-time entertainment
Average day (network downstream) – North America, fixed access (2012)
The Future of the Internet
15
becoming the predominant form of digital distribution. Access
networks experienced significant imbalances (in the order
of 5 to 1) on average between incoming and outgoing traffic
just because the nature of traffic today is media related and
streaming, and therefore mainly flows one way from content
providers to end-users.
Although Internet consumption changes during the day and
peaks between 9 and 10 pm, streaming media and real-time
applications account for the majority of the traffic at any time of
the day.
This evolution of the role and of the type traffic carried through
the Internet pushes the amount of uncontended bandwidth
effectively needed for each individual user to new levels. But
more importantly, it also increases the quality level required for
delivery of content.
As High-Definition media becomes increasingly popular,
reduction of delay and packet loss is becoming critical for the
newer applications. But it becomes essential as a new class of
applications may spread, such as live streaming (e.g. streaming
sport events across popular video platforms) or two-way live
streaming videos (e.g. Skype video-calls and “Hangouts”).
Today, Internet content delivery quality matters and
means money
Nowadays the Internet is a mission-critical platform used by
many companies to sell their own services and products to end-
users. Consequently, and not surprisingly, the quality of the end-
user’s experience has become increasingly important because it
directly affects end-users’ purchasing decisions, with financial
implications that can be quantified.
For example, variation of delay of just 16 milliseconds (on top
of the foreseen average delay of 100 milliseconds) can damage
the end-users’ experience while watching a video stream, and
therefore undermine his willingness to buy video content on
Figure 8: Impact of delay on Media Customer Experience and on an eCommerce site performance
Source: Akamai, ReissRomoli-Telecom Italia, Arthur D. Little analysis
http://www.bth.se/com/mscee.nsf/att...elay_Variation_on_QoE_in_Video_Streaming_pdf/$file/5297_Effect_of_Delay_Delay_Variation_on_QoE_
in_Video_Streaming.pdf
eCommerce conversion rate and web page load time distribution
0
2
4
6
8
10
12
14
16
18
>15sec
Over the last decades, the nature of Internet traffic has changed from static data & text to interactive media content, effectively
transforming the Internet into a new media platform as its usage shifted to richer types of content, particularly streaming video.
The future development of the Internet as a media platform is nowadays challenged by increasing global connectivity, proliferation
of smart devices and streaming media services, which cause spectacularly higher traffic volumes, greater traffic imbalances and
changing traffic patterns. Internet traffic doubles almost every two years, and traffic patterns have changed as a result of real-time
streaming overtaking peer-to-peer as the predominant form of digital distribution. Internet access networks experience significant
in the order of 5:1 imbalances between incoming and outgoing traffic because of the media-related nature of traffic, which mainly
flows one way, from content providers to end-users.
In 2014, the Internet reached over 2.7 billion individuals and has become mission critical for most Content and Application Providers.
Indeed, minor disturbances in the quality of delivery directly impact the willingness of end-users and advertisers to pay for online
services.
Innovation and Investment in IP Interconnection
The Future of the Internet
May 2014
Content
Foreword 3
Key Messages 5
Executive Summary 6
1. The Internet is Vital and Continuously Mutating 11
1.1. The Internet is now mission critical commercially and calls for quality delivery 11
1.2. The IP Interconnection evolves and follows the evolution of the Internet 16
1.3. Introducing the actors in the IP Interconnection value chain 19
1.4. Description of IP Interconnection business models 21
2. So far, the IP Interconnection Value Chain has Adapted Well 31
2.1. Content and Application Providers and Terminating ISPs are setting the pace of
IP Interconnection innovation 31
2.2. Internet Content and Application Providers look for quality control 33
2.3. A new power emerged: the arrival of the Internet Global CAPs 35
2.4. Investment strategies for future IP Interconnection 36
2.5. Friction occurs in the IP Interconnection value chain, but is (so far) quickly resolved 38
3. Revolutionary Future Applications that Require New Delivery Features 40
3.1. The future Internet driven by Internet of Things and Internet of Humans? 40
3.2. Today’s Internet is a Best Effort and finite (yet-not-scarce) resource 43
3.3. New requirements are emerging, beyond bandwidth 48
3.4. New IP Interconnection business models are being developed 52
4. New Business Models Could Accelerate Innovation & Value Creation 55
4.1. The most advanced application landscapes could generate substantial economic value creation 55
4.2. The acceleration of the most advanced application landscapes require to guarantee a link
with the most suitable Internet platform option 56
5. Three Core Assumptions can Drive the Evolution of the Future Internet Platform 59
5.1. The future Internet Platform will, as always, organically grow from stakeholders various interests 59
5.2. Three core assumptions have the potential to shape the future Internet platform 60
5.3. Looking forward, three Internet platform options can be foreseen:
Best Effort 2.0, Quality-Guaranteed Services, Both Worlds 63
Glossary 67
3
Scope of this study: “The Internet” and “IP Interconnection”
The future of the Internet is a widely debated public policy theme all over the world. Questions are raised on
how to preserve the public “best-effort” Internet as an “open” platform for innovation and competition, and
how to combine economic- & societal value creation and sustainable returns on investment. Although so far
the history of the Internet has been an incredible success in organically developing a self-adapting complex
of network business relations, concerns are raised about how the Internet will be able to sustain an adequate
quality of experience for the end-user in the future. This ability may come under pressure by a spectacular
boom in Internet traffic volumes in the coming years, resulting in unprecedented demand for reliable,
ubiquitous Internet access and mass uptake of bandwidth-intensive services and applications. To illustrate
this point: by 2020, more than 50% of the world’s population will be online. This means an increase from 2.7
billion users in 2014 to 5.0 billion users by 2020. By 2025, “The Internet of Things” will comprise around 50
billion connected devices. By 2030, machine-to-machine (“M2M”) communication is expected to constitute
more than 50% of IP traffic.
The question therefore seems justified as to whether the Internet can cope with this evolution, and
who and what is needed for the Internet to evolve and adjust to these changing circumstances.
One part of the answer lies in capacity, quality and traffic management in the Internet access network (fixed
or mobile), or so-called “last mile”, owned or operated by an Internet Access Provider over which end-users
access the Internet. This part is the subject of “net neutrality” discussions and mainly covers the front end,
consumer-facing side of the Internet.
The other part of the answer lies in the so-called “up-stream” side of the Internet. This is where the
Internet access networks connect with (i) each other, (ii) bulk IP traffic transportation networks and undersea
cables connecting continents and (iii) content & application server parks located across the globe. This “IP
Interconnection” part of the Internet solely consists of wholesale agreements, which determine the technical
& economic conditions under which IP traffic is delivered from the originating party (for example, a content
& application provider or an ISP). This is done via several exchangeable delivery networks of multiple Internet
connectivity providers (often used in parallel) to the residential Internet access networks of terminating ISPs,
and vice versa.
IP Interconnection is, and has been, an essential building block for the quality and functionality of the Internet
as ultimately experienced by the end-user, despite the fact that the end-user is no party to IP-Interconnection
arrangements. IP Interconnection models have adapted to changes in Internet usage and traffic patterns
caused by disruptive applications or technologies (for example, predominantly digital distribution technology
moving from decentralized peer-to-peer to centralized streaming) and facilitated IP content delivery
accordingly. In many ways, the extent to which the IP Interconnection sector is able to innovate itself defines
the scope of evolution of the Internet as a platform for future applications.
Foreword
Therefore, IP Interconnection developments have an impact on overarching objectives in Internet public policy debates that,
essentially, focus on warranting end-users’ quality of experience over the public Internet, or “Best-Effort” Internet.
With this report, we want to unravel some of the complexity in IP Interconnection and identify the main drivers of change in IP
Interconnection. We also analyze the effects of investment and innovation by the IP Interconnection players on the future capability
of the public Internet, end-user quality of experience, competition and the scope for new Internet applications.
In order to achieve this, the study first analyzes the latest stage in the evolution of the Internet, and then the corresponding
trends in IP Interconnection reflecting this evolution, asking if the IP Interconnection players will be able to continue to reach
innovative interconnection business models and participate in the Internet value chain. Finally, the study analyzes how different IP
Interconnection scenarios may affect the future of the public “Best-Effort” Internet, which is linked to unlocking the potential of
innovation and new, exciting application paradigms such as the “Internet of Things” and the “Internet of Humans”.
Sincerely,
Gregory Pankert
Partner
TIME* practice
Arthur D. Little
Andrea Faggiano
Principal, Head of SASCAR**
TIME* practice
Arthur D. Little
Karim Taga
Global Practice Head
TIME* practice
Arthur D. Little
* TIME: Telecommunication, Information Technology, Media and Electronics
**SASCAR: Strategic Advisory Services for Competition and Regulation
The Future of the Internet
5
The Internet is vital, continuously evolving and has developed into a new media platform
1. The Internet has been transformed into a new media platform, as the nature of Internet traffc has changed from static data &
text fle transfer to streaming interactive media content.
2. The Internet has become mission critical for most Content & Application Providers. Minor disturbances in the quality of delivery
directly impact the willingness of end-users and advertisers to pay for online services.
3. The future development of the Internet as a media platform is impacted by increasing global connectivity, proliferation of smart
devices and streaming media services which cause spectacularly higher traffc volumes, greater imbalances in traffc fow and
changing traffc patterns.
4. IP Interconnection is an essential building block for the quality & functionality of the Internet as ultimately experienced by the
end-user, despite the fact that the end-user is no party to business-to-business IP Interconnection arrangements.
IP Interconnection, so far, adapted well to support the changing nature of the Internet, and remains dynamic and
competitive
5. The IP Interconnection value chain converges, but remains dynamic and competitive. Proliferation of Content Delivery Networks
and Internet Exchanges, commoditization of IP transit and CDN prices challenge existing interconnection models and enable
new ones.
6. From the early days of “IP transit” and “Peering”, a genuine mix of viable application/content delivery strategies is accessible to
all players seeking connectivity.
7. Content & Application Providers and ISPs are setting the pace and determining the nature of IP Interconnection innovation by
vertically integrating and interconnecting directly, which disintermediates pure Internet connectivity providers to some extent.
8. Changes in the IP Interconnection ecosystem lead to tension between IP Interconnection players. However, disputes concern
less than 1% of all IP Interconnection agreements and are solved without regulatory intervention in more than half of these
cases.
9. End-users have not been substantially or structurally affected by IP Interconnection disputes.
Future applications require IP Interconnection models to evolve towards providing higher-quality assurances,
which will impact the current “best-effort” Internet
10. Innovation in IP Interconnection is needed to support further development of the Internet and accelerate take-up of next-
generation applications (Internet of Things, Internet of Humans) that require IP Interconnection Quality of Service (latency, jitter,
packet loss) extended with new parameters (e.g. security, data protection).
11. Variants of Paid Peering, Deep Caching, Assured Delivery or Secure M2M are among the innovative IP Interconnection business
models that could lay the foundation for an advanced Internet platform, based on assured end-to-end Quality of Service Internet
Platform – complementary to “Best Effort”.
12. “Best-effort” Internet is and will no doubt continue to be essential in the future, and there is early evidence to indicate that it can
continue to improve and coexist with complementary end-to-end Quality of Service platforms if properly monitored.
13. Private investment in IP Interconnection has led to structurally improved conditions for the future development of the public
Internet. Content comes closer to end-users (by direct interconnection and local content caching), Internet performance is
improved by adoption of new application technologies (e.g. “adaptive streaming”) and IP network resources are abundant (e.g.
higher capacity in the “last mile”).
Key Messages
The Future of the Internet
6
The Internet is vital, continuously evolving and has developed into a new media platform
Over the last decades, the nature of Internet traffic has changed from static data & text to interactive media content, effectively
transforming the Internet into a new media platform as its usage shifted to richer types of content, particularly streaming video.
The future development of the Internet as a media platform is nowadays challenged by increasing global connectivity, proliferation
of smart devices and streaming media services, which cause spectacularly higher traffic volumes, greater traffic imbalances and
changing traffic patterns. Internet traffic doubles almost every two years, and traffic patterns have changed as a result of real-time
streaming overtaking peer-to-peer as the predominant form of digital distribution. Internet access networks experience significant
in the order of 5:1 imbalances between incoming and outgoing traffic because of the media-related nature of traffic, which mainly
flows one way, from content providers to end-users.
In 2014, the Internet reached over 2.7 billion individuals and has become mission critical for most Content and Application Providers.
Indeed, minor disturbances in the quality of delivery directly impact the willingness of end-users and advertisers to pay for online
services.
IP Interconnection, so far, adapted well to support the changing nature of the Internet, and remains dynamic
and competitive
IP Interconnection is an essential building block for the quality & functionality of the Internet as ultimately experienced by the end-
user, despite the fact that the end-user is no party to business-to-business IP-Interconnection arrangements. IP-Interconnection
models adapt to changes in Internet traffic patterns caused by disruptive applications, or technologies, and facilitate IP content
delivery accordingly. In many ways, the extent to which the IP Interconnection sector is able to innovate itself defines the scope of
evolution of the Internet as a platform for future applications.
Executive Summary
Figure A - The changing nature of the Internet and the new capacity/quality requirements
* downstream traffic
COMMUNICATION: services provided by application (Skype, WhatsApp, iMessage, FaceTime,etc.); DATA: file sharing (Bit Torrent, eDonkey, etc),
web browsing, social networking, email, etc.; MEDIA: streamed or buffered audio and video (Netflix, non-linear TV services;
1. 2009-2012 CAGR; 2. Interviews. Source: ITU, Sandvine; Arthur D. Little analysis
More
bandwidth
More quality
(e.g. reduced delay)
Media
Communication
Data
Peer-to-peer
download
5:1
In/out traffic
unbalance
2
+39%
yearly traffic
growth
1
na
na
21%
10%
33%
11%
33%
14%
Data
Media
(streamed or
buffered one-
way audio and
video)
Communi-
cation
% of fixed traffic at busy hour (North America illustration)
*
*
32%
43%
53%
62%
68%
54%
45%
36%
2011
2% 2%
2010
3%
2009 2012
The Future of the Internet
7
The Internet and the underlying IP Interconnection ecosystem demonstrated an organic ability to evolve and adapt. Alternative
business models (such as peering and Content Delivery Networks) challenged existing ones and improved the overall efficiency of IP
Interconnection, leading to a cost reduction of around 30% per annum over the last decade.
From the early days of “IP transit” and “Peering”, a genuine mix of IP Interconnection models is currently available to both ISPs and
Content & Application Providers (CAPs) seeking connectivity. This is the result of three major developments:
1. Decentralization of the Internet: the emergence of national and regional Internet Exchanges facilitates private peering
arrangements by increasing number of ISPs’ access networks edges in one central location.
2. Commoditization of IP Interconnect prices (falling IP transit, CDN or router costs) led to substitutability of IP Interconnection
products and countervailing powers in the IP Interconnection value chain.
3. Proliferation of Content Delivery Networks: Content & Application Providers leverage the increased value of their Internet
content by building proprietary caching server parks or, alternatively, using independent, commercial CDN services that are
located close to the ISPs’ access networks.
Still, the majority of Internet traffic is progressively being concentrated to a limited number of large Content & Application Providers
and a few wholesale carriers. In 2013, 35 networks carried 50% of all Internet traffic in North America, down from 150 networks in
2009. The concentration of IP traffic is a major evolution in the IP Interconnection value chain, and has the potential to influence the
negotiating power among connectivity stakeholders and affect the current equilibrium in the Internet ecosystem.
In the last years, the largest Content & Application Providers and Internet Service Providers have been setting the pace and
determining the nature of IP Interconnection innovation through vertical integration. Content & Application Providers seek end-
user proximity and are increasingly investing in proprietary Content Delivery Networks or relying on third-party CDNs. ISPs invest
in network-based content delivery platforms (“deep caching”) for internal purposes and as a service to third-party Content &
Application Providers.
As a result, Content & Application Providers and Internet Service Providers increasingly interconnect directly, disintermediating pure
Internet connectivity providers to some extent. Improving control on the quality of delivery over the Internet is the main motivation.
This is true not only for Internet-based CAPs, but also increasingly for the video-streaming strategies of traditional broadcasters
(e.g. BBC iPlayer’s average daily unique users grew 33% year on year since 2009). The equilibrium in the IP Interconnection value
chain has subsequently changed, and traditional IP Interconnection players have adapted to maintain their competitiveness. Internet
Figure B – Traffic concentration trends (2007, 2009, 2013)
* 2013 figures refer to North America Internet traffic
Source: Deepfield; Arthur D. Little analysis
Akamai
40%
20%
0
15
30
45
60
75
0 40 80 120 160 200 240 280 320 360 400
2013*
2009
2007
% of North America traffic
# of networks
50% of traffic from
1000s of networks
50% of traffic from
150 networks
50% of traffic from
35 networks
% of North America traffic
The Future of the Internet
8
connectivity providers such as IP Transit providers, independent CDN providers and Internet Exchanges are under pressure to
innovate and diversify their service offerings (e.g. offering “partial transit”, commercial open CDNs or web security) or attract a critical
mass of traffic through consolidation (e.g. international carrier Level 3 acquiring its competitor, Global Crossing).
Changes in the IP Interconnection ecosystem meant that tensions between IP Interconnection players intensified. However,
disputes concern less than 1% of all IP Interconnection agreements, and are solved without regulatory intervention in more than
half of these cases. There are a number of reasons for this:
n
IP Interconnection (including upgrade) costs account for just a marginal share, i.e. less than 1% of the overall connectivity costs.
n
Countervailing powers emerged by changing the IP Interconnection economics that keep the value chain in balance:
Figure C – Trends over the IP Interconnection value chain
Source: Arthur D. Little analysis
Drivers
Strategic
moves
Large CAPs seek for user proximity
END-
USERs
? Grow scale
? Defend profitability
and investment
payback
? Secure quality of
service for own
applications
? Search for
economies of scale
? Reach/keep critical
mass
? Attract new
members
? Diversify revenues
? Grow scale
? Defend profitability
? Look for new
revenue streams
and monetize
eyeballs
Partnering with
access providers
Partnering with
access providers
own
infrastructure
own
platforms
own
infrastructure
own
platforms
Core Business
own
platforms
Core Business
own
platforms
IP Transit
providers
Content Distribution
providers (CDN)
Terminating ISPs
Interconnection
Exchanges (IX)
Content &
Application
providers
Explosion of Internet
Exchange and Peering
Significant drop in
router prices
Explosion of Content
Delivery Networks
Commoditization of IP Transit
and CDN prices
Router cost per Gbps ($)
CDN video prices ($/GB)
IP Transit prices ($/Mbps)
0
1
2
3
4
5
6
7
8
9
10
11
12 1.20
1.05
0.90
0.75
0.60
0.45
0.30
0.15
0.00
2013E
0.05
1.6
2012 2011 2010 2009 2008
0.30
12.0
-33%
-30%
30
3
122
2010 2008 2006
25
2012
155
Global # of CDN operators
30%
Largest CAPs’ CDN
Commercial CDN
Network Operators’ CDN
Global # of IXs by region
2
0
1
1
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
2
0
0
7
2
0
0
8
2
0
0
9
2
0
1
0
Latin
America
North
America
Europe Asia
0
47.000
1.000
2007 2012 2003 1999
50
100
150
250
300
350
400
Figure D – Economic drivers of IP Interconnect evolution and countervailing powers
Source: ITU, Informa, Packet Clearing House, Dr. Peering, Cisco, streamingmedia.com, Web sites, Arthur D. Little analysis
-24%
The Future of the Internet
9
a) IP Transit and Peering have become substitutes in terms of cost
b) Falling IP Transit and CDN prices make high-quality transport and CDN strategies accessible to smaller CAPs
c) Falling IP Transit prices balance against paid-peering cost pressures
d) Strong retail competition prevents market foreclosure by ISPs
End-users have not been substantially or structurally affected by IP Interconnection disputes. The commercial interest of parties
prevailed, and mutually acceptable solutions were found. Associated Interconnection costs have not proven to be prohibitive to core
business models so far.
Future applications require IP Interconnection models to evolve towards providing higher-quality assurances,
which will impact the current “best-effort” Internet
In addition to continuous improvement of connectivity between Content & Application Providers and access networks, innovation in
IP Interconnection can support further development of the Internet and accelerate the take-up of next-generation applications that
require uncompromised quality. In particular, the Internet of Things and the Internet of Humans application landscapes can unlock
an economic value potential in the range of trillions of euros by 2020. However, advanced Internet platforms, i.e. beyond Best-Effort,
may be required for next-generation applications, which could bring an Internet of Things and an Internet of Humans to life.
IP Interconnection Quality of Service needs to be extended to new parameters (e.g. latency, jitter, packet loss, security, and
data protection). As the Internet evolves from nice-to-have services to mission-critical services, next-generation applications for
sectors such as the Financial Services industry, the Electronic Payment sector, high-security Governmental Bodies (police, military,
emergency services, etc.) will generate a demand for new IP Interconnection requirements going well beyond additional throughput
capacity. It will expand to delivery features relevant for streaming video, such latency reduction, availability, jitter control and packet-
loss limitation. Security and data protection deserve special attention as they play a critical role in the safe use of next-generation
applications, especially in scenarios foreseeing the Internet of Things leading to M2M applications such as connected cars with
remote-start features.
Variants of Paid Peering, Deep Caching, Assured Delivery and Secure M2M are among the innovative IP Interconnection business
models that could lay the foundation for an advanced Internet platform, based on an assured end-to-end Quality of Service Internet
platform – complementary to Best-Effort.
Figure E – Innovation in IP Interconnect business models
Source: Arthur D. Little analysis
Best-Effort High-Quality
Openness
Offered service
Assured quality
Reporting services
Interconnection
point
Application
risk sharing
None
Selected networks
Secure
networks
Transfer +
close caching
None
None
Co-design
risk sharing
Commercial launch
risk sharing
Peering IP Transit Deep Caching
new
Paid Peering Assured Delivery
new
Secure M2M
new
IP Interconnection
dimensions
Entry IP port
(upstream )
All networks
Transfer
extra threshold
Balanced
transfer
Caching +
Transfer
IP port in the Local Exchange
(downstream, closer to users)
Guarantee
on port
availability
Guarantee on
access transfer
(delay, jitter)
Guarantee on
transfer (by traffic
classes)
Reporting QoS between
IP port and access
gateway
Reporting QoS
between IP port
and user device
Reporting on
QoS at IP port
Transfer +
hosting locally
The Future of the Internet
10
Still in the advent of new IP Interconnection business models, the Best-Effort Internet is, and will no doubt, continue to be essential
in the future, and there is early evidence to indicate that it can continue to improve and co-exist with complementary end-to-end
quality of service platforms if properly monitored. Best-Effort has long co-existed with business-to-business IP managed services,
as well as with ISPs’ managed IPTV platforms, and its average and peak connection speeds still increased by respectively 12% and
23% since 2007, increasing to 21% and 26% since 2011.
Private investments in IP Interconnection resulted in a number of trends with structurally improved conditions for the future
development of the public Internet:
a) Content comes closer to end-users: Direct interconnection with fewer Internet connectivity providers allows for better
structural conditions on Quality of Service - i.e. lower latency, lower risk of packet loss and jitter.
b) New application technologies improve performance: This includes codecs, adaptive streaming and content distribution
algorithms.
c) Abundance of IP network resources: Capacity in the “last mile” became larger. IP transit and Content Delivery services
were being commoditized, and new quality delivery opportunities were created, with network-based “deep-caching”
technologies.
d) High dynamism in the Internet ecosystem: The high value at stake for all stakeholders allows new disputes to be resolved
quickly and create new business relationships.
The public Internet will stand to benefit mostly from private investments in IP Interconnect architecture aimed at shortening the
distance that Internet traffic needs to travel before it reaches the “last-mile” Internet access networks. This is accomplished by
storing popular content/applications in local servers that form part of proprietary or commercial CDNs, or in network-based “deep-
caching” servers. Shorter travel distances for IP content with fewer intermediaries implies less chances for “bumps” in the road.
It also increases the prospects for more manageable end-to-end controls, leading to an overall higher quality of experience for the
end-user.
Figure F – IP Interconnection structurally contributing to Internet’s improvement
Source: Akamai, Arthur D. Little analysis
IP Interconnection trends:
better structural conditions
Best-Effort Internet is continuously improving
? Content comes closer to end-users
(direct interconnection with fewer intermediaries, allowing
for better structural conditions on QoS: lower latency, lower
risk of packet loss and jitter)
? New application technologies improve performance
(codecs, adaptive streaming, content distribution
algorithms)
? Abundance of IP network resources
(larger pipes in the last mile, IP Transit and Content Delivery
services being commoditized, new opportunities with deep
caching technologies)
? High dynamism in the ecosystem
(the high value at stake for all players allows the new
disputes to be resolved rapidly and new business
relationships to be created)
0
1
2
3
4
5
6
7
8
Q2’08 Q4’07
+21% p.a.
+12% p.a.
Q4’13 Q2’13 Q4’12 Q2’12 Q4’11 Q2’11 Q4’10 Q2’10 Q4’09 Q2’09 Q4’08
AMERICAS EMEA GLOBAL
Average connection speed
The Future of the Internet
11
1.1. The Internet is now mission critical
commercially, and calls for quality delivery
The Internet rapidly and dynamically evolved from an
experimental network to a mass-market interaction
platform
The story of the Internet began in 1957, when the USA
responded to the USSR, launching Sputnik into space with the
creation of the Advanced Research Projects Agency (ARPA). The
Agency’s mission was to become the leading force in science
and new technologies. Around 1994 the Internet was used for
the first time as an open commercial platform, enabling the
launch of the first Internet ordering system (Pizza Hut) and the
first Internet Bank (First Virtual). From that point on, all the major
Internet companies emerged: Google was launched in 1998,
MySpace in 1999, Apple’s iTunes Store in 2003, Facebook in
2004, YouTube in 2005 and Twitter in 2006. Netflix (created
in 1998) started to offer online streaming services in 2008.
Soon, their rapid growth and potential success would bring
these Internet companies to become listed companies, Google
showing the lead in 2004. Facebook (2012) and Twitter (2013)
made remarkable entries to the stock market.
From its birth in 1969 to nowadays, the Internet has evolved and
mutated in many different ways:
n
Internet usage ceased to be a US-centric phenomenon and
achieved a global reach;
n
Traffic boomed and regional poles emerged, while the nature
of Internet content progressively changed from static text
and simple data to interactive media and entertainment;
n
The way of accessing the Internet shifted from dialing in via
fixed networks to an always-on mobile experience
n
New network requirements emerged, and quality of
delivery – which, in the early stages was not that important –
became mission critical.
The popularity of the Internet grew substantially, and the
penetration of Internet users within the world’s population
strongly increased.
Over the 2005-2013 period, the number of individuals using the
Internet
1
grew yearly at 13%, from 1 billion to almost 3 billion,
mainly outside the US.
1 Source: ITU; an Internet user is someone aged 2 years old and above who
went online in the past 30 days
1. The Internet is Vital and Continuously
Mutating
Figure 1: The Internet timeline
Commercial age: 25 years, so far Research age: 30+ years
1960 1965 1980 1985 1995 2010
1957: The USA creates the
Advanced Research Projects
Agency (ARPA)
1969: First switched
network (called ARPANET)
is created, connecting four
different nodes in
California and Utah
1975
1989: ARPANET
ceases to exist
1993: the first web browser,
Mosaic (later Netscape) is
released
1990
1994:
? First Internet ordering system
(pizza Hut) is launched
? First online payment system
(First Virtual) is launched
2005: YouTube is
launched
2005
1998:
Google is
launched
1999: MySpace
is launched
2003: Apple launches
iTunes store
1970: NCP protocol is
developed as
ARPANET’s host-to-
host protocol
1973: First written
version of the TCP/IP
is developed
1976: SATNET, a satellite
program, is developed to link
the USA to Europe
1970
2005:
Facebook is
launched
1999:
Wi-Fi is
standardized
2012: Facebook
becomes a listed
company
2000
1983: Domain
Name System
is introduced
1983: TCP/IP
becomes the
standard
internet
protocol
2008: Netflix starts
to offer online
streaming services
1972: Ethernet
is established by
Metcalfe
1972: First program
devoted to email is
created
1989: CERN releases the
World Wide Web
1990: The first search engine
is created, called the Archie
Search Engine
1996: Nokia
releases the
first cell phone
with Internet
access
2006: Twitter is
launched
2013: Twitter
becomes a listed
company
2004: Google
becomes a listed
company
Source: Arthur D. Little
The Future of the Internet
12
Regional data from ITU reveals that this growth is consistent
across all world regions. Europe’s and America’s relatively low
growth during these recent years (7% and 8%, respectively) is
explained by their earlier take-up of the Internet, and both regions
together now account for only 38% of global Internet users.
However, Internet traffic origination is not equally distributed
across global regions: most of the IP traffic originates in North
America (34%), followed by Asia & Pacific with 33%, Europe
with 24%, Latin America with 8% and the Middle East & Africa
with only 2%.
CISCO indicates that IP traffic will continue to grow, but at lower
growth rates: highest growth rates are expected from 2013 to
2017 in the Middle East & Africa (35%) and Asia & Pacific (24%).
In the beginning, Internet access relied mainly on fixed-network
infrastructure and end-users experienced the Internet through
desktop computers. Over the years, a vast array of connected
devices emerged, such as smartphones and tablets, through
which end users could experience the Internet in mobility.
These events significantly increased Internet’s accessibility and
boosted its penetration in the global population.
The number of broadband (BB) subscriptions has grown over
2007-2013 at a growth rate of 28.7 %, from 614 million to 2.8
billion, but when looking closer at their composition, it is clear
how the mobile broadband development did become the main
driver of this significant growth trend, especially in developing
countries
2
.
Between 2007 and 2013, the share of mobile broadband access
increased from 43% of total broadband subscription to 75%.
As the number of Internet users and broadband subscriptions
has increased, the volumes of IP traffic carried over the Internet
2 Classifcation of developing and developed country is available athttp://www.
itu.int/ITU-D/ict/defnitions/regions/index.html, and has been made according
to UN M49
Figure 2: Individuals using the Internet
Source: ITU, Arthur D. Little analysis
44%
2013
2.7
46%
21%
17%
5%
22%
19%
5%
5%
4%
2010
2.0
43%
23%
21%
5%
5%
4%
2009
1.7
42%
25%
22%
4%
4%
4%
2008
5%
40%
26%
24%
4% 4%
3%
2007
1.3
37%
29%
25%
3% 3%
2%
2006
1.1
35%
31%
26%
3%
3%
5%
2005
1.0
34%
31%
27%
3%
3%
2%
5%
1.5
2012
2.5
45%
22%
18%
5%
5%
2011
2.3
2%
Individuals using the Internet (Bln) CAGR
Europe +7%
Asia & Pacific +18%
The Americas +8%
CIS +23%
Middle East +23%
Africa +30%
Figure 3: Global IP traffic by region
2017E 2012 2016E 2015E 2014E 2013E
120.643
14.831
101.055
83.837
10.174
55.553
43.570
2011 2010 2009 2008
68.892
28.046
20.182
Global Internet traffic (PB/month) CAGR
(’08-’13)
+57%
+70%
+36%
+33%
+49%
North America
Asia Pacific
Latin America
Middle East and Africa
Europe
Source: CISCO, Arthur D. Little analysis
The Future of the Internet
13
have grown tremendously. Cisco Visual Networking Index shows
that traffic has risen from 10.1 exabytes (1018 bytes, or 1 billion
gigabytes) per month in 2007 to 55.5 EB/month in 2013, with a
40% CAGR, and is expect to reach more than 120 EB/month in
2017 (with a growth rate of 21% from 2013 to 2017).
Fixed-Internet traffic accounts for about 70% of Global IP traffic,
and has grown at a CAGR of 37%, reaching 39 EB/month. It is
expected to reach 82 EB/month in 2017, counting for roughly
68% of the total IP traffic.
Managed IP services such as IP Virtual Private Networks (IP
VPN), historically applied in business-to-business environments,
have also grown significantly from 2007, at a CAGR of 49%.
These have reached 15 EB/month in 2013, accounting for 26%
of total IP traffic; CISCO projections in 2013 indicate that IP
managed services traffic could reach 27 EB/month by 2017,
slightly decreasing its share over total Internet traffic (23%).
Global mobile traffic has grown at notable rates (CAGR of 117%),
reaching 1.6 EB/month in 2013, but it still accounts for only 3%
of total IP traffic; however, CISCO expects it to rise to 11 EB/
month by 2017, accounting for more than 9% of total IP traffic.
This growth of IP traffic volumes – data consumption per user
grew from 12 gigabytes per user per month in 2008 to roughly
Figure 4: Internet Broadband Subscription and Global IP traffic by type
Source: ITU, Arthur D. Little analysis
411
468
527
588
638
422
615
778
1.556
696 Fixed broadband
Mobile broadband
2013E
2.792
2.096
2012
2.194
2011
1.743
1.155
2010
1.305
2009
1.083
2008
833
Broadband subscriptions (mln)
+12%
+41%
Fixed Internet
2013E
Managed IP
Mobile data
55.552
2012
43.570
2011
28.035
2010
20.181
2009
14.832
2008
10.174
Global Internet traffic (PB/month) CAGR
(’08-’13)
+117%
+49%
+37%
CAGR
(’08-’13)
Figure 5: The changing nature of the Internet and the new requirements
* downstream traffic
COMMUNICATION: services provided by application (Skype, WhatsApp, iMessage, FaceTime,etc.); DATA: file sharing (Bit Torrent, eDonkey, etc),
web browsing, social networking, email, etc.; MEDIA: streamed or buffered audio and video (Netflix, non-linear TV services;
1. 2009-2012 CAGR; 2. Interviews. Source: ITU, Sandvine; Arthur D. Little analysis
More
bandwidth
More quality
(e.g. reduced delay)
Media
Communication
Data
Peer-to-peer
download
5:1
In/out traffic
unbalance
2
+39%
yearly traffic
growth
1
na
na
21%
10%
33%
11%
33%
14%
Data
Media
(streamed or
buffered one-
way audio and
video)
Communi-
cation
% of fixed traffic at busy hour (North America illustration)
*
*
32%
43%
53%
62%
68%
54%
45%
36%
2011
2% 2%
2010
3%
2009 2012
The Future of the Internet
14
20 gigabytes per user per month in 2013 – is not only driven
by the increase in the number of broadband subscribers, but is
also, and most importantly, linked to the change in nature of the
Internet traffic itself.
The nature of Internet traffc changed from static data
and text to interactive media content
The latest measures available show that the bulk of total usage
growth comes from real-time streaming devices; the Internet
has transformed itself from a data- and file-transfer platform into
a new-media platform, and its usage has shifted to richer types
of content, particularly video.
Today more than 60% of Internet traffic in the US is media
related and, seemingly, such share is expected to grow further
in the coming years
The consequences of such a shift materialized into increased
demand for a higher bit rate and delivery quality. During the
early ages of the Internet, communication interactions were
established through sequential (asynchronous) applications, and
on-time delivery was not important (e.g. emails); nowadays,
higher throughput and reduced delivery time are essential for a
good quality of experience (see figure 5 overleaf).
Furthermore, traffic kept doubling almost every two years,
and traffic patterns changed as a result of real-time streaming
Figure 7: New connectivity requirements
Source: Arthur D. Little analysis
Uncontended capacity
More quality
(e.g. reduced delay or packet losses)
P2P applications
Near-real-time
streaming
Download Streaming
One-way
live streaming
Live football
matches
Prime-time events
Two-way
live streaming
HD
Ultra HD
HD
HD
Figure 6: Daily Internet traffic by content type
Note: Storage and Back-up Services: PDBox, Netfolder, Rapidshare, etc..; Bulk Entertainment: iTunes, movie download services; P2P Filesharing: BitTorrent, eDonkey, Gnutella,
Ares, etc…; web Browsing: HTTP, WAP browsing; Real-time Entertainment: streamed or buffered audio and video, peercasting, specific streaming sites service (Netflix, Hulu,
YouTube, Spotify,..);
Source: Sandvine, Arthur D. Little analysis
? 60% of Internet traffic consists of
video (real-time entertainment +
bulk entertainment) during most of
the day
? During night hours Internet traffic
reaches its peak
Outside top 5
Storage and Backup services
Bulk entertainment
P2P file sharing
Web browsing
Real-time entertainment
Average day (network downstream) – North America, fixed access (2012)
The Future of the Internet
15
becoming the predominant form of digital distribution. Access
networks experienced significant imbalances (in the order
of 5 to 1) on average between incoming and outgoing traffic
just because the nature of traffic today is media related and
streaming, and therefore mainly flows one way from content
providers to end-users.
Although Internet consumption changes during the day and
peaks between 9 and 10 pm, streaming media and real-time
applications account for the majority of the traffic at any time of
the day.
This evolution of the role and of the type traffic carried through
the Internet pushes the amount of uncontended bandwidth
effectively needed for each individual user to new levels. But
more importantly, it also increases the quality level required for
delivery of content.
As High-Definition media becomes increasingly popular,
reduction of delay and packet loss is becoming critical for the
newer applications. But it becomes essential as a new class of
applications may spread, such as live streaming (e.g. streaming
sport events across popular video platforms) or two-way live
streaming videos (e.g. Skype video-calls and “Hangouts”).
Today, Internet content delivery quality matters and
means money
Nowadays the Internet is a mission-critical platform used by
many companies to sell their own services and products to end-
users. Consequently, and not surprisingly, the quality of the end-
user’s experience has become increasingly important because it
directly affects end-users’ purchasing decisions, with financial
implications that can be quantified.
For example, variation of delay of just 16 milliseconds (on top
of the foreseen average delay of 100 milliseconds) can damage
the end-users’ experience while watching a video stream, and
therefore undermine his willingness to buy video content on
Figure 8: Impact of delay on Media Customer Experience and on an eCommerce site performance
Source: Akamai, ReissRomoli-Telecom Italia, Arthur D. Little analysis
http://www.bth.se/com/mscee.nsf/att...elay_Variation_on_QoE_in_Video_Streaming_pdf/$file/5297_Effect_of_Delay_Delay_Variation_on_QoE_
in_Video_Streaming.pdf
eCommerce conversion rate and web page load time distribution
0
2
4
6
8
10
12
14
16
18
>15sec