A Busy Person's Guide to How the

Internet Works (and is paid for)

12 July 2022

Mark McFadden, Aude Schoentgen, Karim Bensassi-Nour

plumconsulting.co.uk
About Plum
Plum offers strategy, policy and regulatory advice on
telecoms, spectrum, online and audio-visual media issues.
We draw on economics and engineering, our knowledge of
the sector and our clients’ understanding and perspective to
shape and respond to convergence.

About this study

This is a report for Google that addresses the topic: How the
Internet Works (and is paid for) – a 2020s refresh (to inform
policymaking).

This is the first of three parts of the study: a concise

introduction to the way the Internet works.

Plum Consulting
10 Fitzroy Square
London
W1T 5HP

T +44 20 7047 1919

E [email protected]
A Busy Person's Guide to How the Internet Works (and is paid for)

Contents
1 Introduction 4

2 How Do You Get Access to the Internet? 5

3 What Does Information Look Like on the Internet? 8

4 How is the Internet Coordinated and Governed? 10

5 How Does Information Move Around the Internet? 11

6 An Example of How Content is Delivered 13

7 How is Peering and Transit Paid For? 14

8 What are Internet eXchange Points and Content Delivery Networks? 16

9 What is Content on the Internet? 17

10 Who Pays for Content? 18

11 The Internet and Regulation 19

12 Advantages of the Internet’s Model 20

© 2022 Plum Consulting 3

A Busy Person's Guide to How the Internet Works (and is paid for) 1 Introduction

1 Introduction
The Internet is a collection of people, nodes, networks and content. Often it is described as a cloud made up of
connections between devices, but this is misleading because there are rarely direct connections over the
Internet. In fact, the Internet is completely decentralised with many nodes and combinations of direct and
indirect connections between them.

In the early days of the Internet, people used computers to construct and send requests to other computers that
serviced those requests. This request and response model, over a collection of decentralised networks served
the Internet very well in its early decades. Today’s Internet is far more complex. We use the term “nodes”
because the Internet connects far more than computers. Today’s Internet connects things, services, mobile
devices, cars and much more. Nodes are simply devices on networks that can send or receive information.

In its simplest form, each node has an address1 that can be used as a locator to which other nodes can send
requests or information. The address allows the Internet to deliver the message to the correct destination even
though the message usually passes through many other intervening nodes.

There are special nodes in the network that connect diverse networks to each other. These “routers” have the
job of receiving messages and forwarding them on toward their intended destination. The routers use the
address on the message as a tool to decide how to send the message onward, and closer, to its final
destination.

1
An Internet Protocol address, or IP address for short.

© 2022 Plum Consulting 4

A Busy Person's Guide to How the Internet Works (and is paid for) 2 How Do You Get Access to the Internet?

2 How Do You Get Access to the Internet?

Consumers and enterprises
both gain access to the
Internet through Internet
Service Providers (ISPs). ISPs
sell access to the Internet as a
service. For a consumer, this
often means one of two
possibilities: first, a fixed
Internet access service in the
home; and second, Internet
access for a mobile device. In
the first case the ISP often
provides a router or home hub
that allows devices in the home
to connect. Messages from
nodes inside the home are
directed to the home router
and then out to the Internet
through the ISPs own network.
Enterprises and businesses
usually have connections with
more capacity. Still, the basic
picture is often the same: the
enterprise has a router on the
“edge” of its internal network
and nodes inside the network
send messages through that
router and out to the ISP’s
network connections.

Figure 2.1: Connected

devices of the Internet

© 2022 Plum Consulting 5

A Busy Person's Guide to How the Internet Works (and is paid for) 2 How Do You Get Access to the Internet?

Figure 2.2: Information Flows on the Internet

Source: Plum

One of the most significant ways that the Internet has changed is how people connect to their ISPs. Many years
ago, most connections used the copper wires that also provided traditional voice communications. Today, the

© 2022 Plum Consulting 6

A Busy Person's Guide to How the Internet Works (and is paid for) 2 How Do You Get Access to the Internet?

Internet uses a variety of access technologies that allow people and businesses to connect, including cable
services, cellular and mobile Internet, fibre optic, fixed wireless and even access to the Internet via satellites.

ISPs get paid by end-users (either consumers or enterprises) for Internet access provision. End-users are
generally charged on an unmetered basis through “flat rates” or on a usage basis. The flat rate for Internet
access includes the Internet connection itself and the transfer of an unlimited volume of data. In the early days
of the Internet, before the advent of broadband, most customers’ monthly Internet access spending had two
components: an Internet service subscription fee, and a metered fee related to the amount of time, or amount
of traffic, a customer used while connected to the Internet.

In terms of end-user prices for Internet access, there is great variation across different regions of the world.
These disparities can be explained by the differences in regulatory regimes, levels of competition, infrastructure
development and consumer bases.

© 2022 Plum Consulting 7

A Busy Person's Guide to How the Internet Works (and is paid for) 3 What Does Information Look Like on the Internet?

3 What Does Information Look Like on the

Internet?
The basic unit of information on the Internet is called a “packet.” Every time a node sends messages on the
Internet, it constructs a series of packets that function as envelopes containing the information being
exchanged. The packets usually have two parts: a “header” that contains information about the packet, and the
“payload” which contains the content of the message.

Packets are simply made up of ones and zeroes, known as “bits.” All messages traverse the Internet in this binary
format. The nodes in the network are responsible for encoding information (email, pictures, video, etc.) from its
original format into bits that can be assembled into packets.

One of the most important parts of the header is the addressing information it contains. Each packet’s header
has the address of the node the message is being sent to, as well as the address the message is being sent from.
Routers in the Internet use these addresses to decide how to forward the packet on and closer to its intended
destination.

The payload of the message contains the content sent between nodes. For example, a picture would be made
up of many packets – each to be sent between the sending and receiving node. The sending node would break
the image into multiple packets and send all of them, through the routers toward the receiving node. The
destination would then take the packets and reassemble them into the original picture. Most things sent on the
Internet are too big to fit in a single packet, so this disassembly and assembly process takes place for almost
every service or application on the Internet.

Figure 3.1: : Sending a Movie Across the Internet

© 2022 Plum Consulting 8

A Busy Person's Guide to How the Internet Works (and is paid for) 3 What Does Information Look Like on the Internet?

It's crucial that both the sender and receiver use the same approach for disassembly and assembly of
information. To ensure that the sender and receiver can understand each other, a set of standard, interoperable
“protocols” have been developed so that all nodes will have a common set of rules for exchanging messages
and information. These protocols are simply rules for nodes to use to talk to each other. Very famous protocols
include the Hypertext Transport Protocol (HTTP), which allows the exchange of messages for the World Wide
Web and the Simple Mail Transport Protocol (SMTP), which allows for the transport of messages for electronic
mail.

Applications as complex as video conferencing and as simple as Internet chat all have this defining characteristic
in common: the use of standardised protocols to exchange information contained in packets over the networks
connected to the Internet.

© 2022 Plum Consulting 9

A Busy Person's Guide to How the Internet Works (and is paid for) 4 How is the Internet Coordinated and Governed?

4 How is the Internet Coordinated and

Governed?
The protocols that make the Internet possible are developed by a diverse group of international, technical
organisations – sometimes called “standards bodies” or “standards development organisations” (SDOs),
including:

● The Internet Engineering Task Force (IETF) is the primary SDO for establishing Internet communication
and transport protocols.

● The Institute of Electrical and Electronics Engineers (IEEE) is the key SDO for describing the rules for
sending bits over wires or wireless connections.

● The International Telecommunications Union’s Standardization Sector (ITU-T) is responsible for

developing and publishing telecommunications protocols.

● The International Organization for Standardization (ISO) defines standards for application in a very
diverse group of sectors, including technology, business, government, and society.

● The World Wide Web Consortium (W3C) creates the rules for exchanging information on the World
Wide Web.

● The Internet Corporation for Assigned Names and Numbers (ICANN) is responsible for ensuring
the stable and secure operation of the Internet’s unique identifier systems, by coordinating the
management of the technical elements of the Domain Name and numbering System (DNS) to ensure
universal resolvability so that all users of the Internet can find all valid addresses.

● Regional Internet Registries (RIRs) such as RIPE and APnic coordinate the Internet’s unique numbering
and addressing spaces. The RIRs work through a process of consensus for their regional community.

© 2022 Plum Consulting 10

A Busy Person's Guide to How the Internet Works (and is paid for) 5 How Does Information Move Around the Internet?

5 How Does Information Move Around the

Internet?
The Internet is not one, big, unified network, it is sometimes called a “network of networks.” The networks that
make up the building blocks of the Internet are called Autonomous Systems (AS), and there are about 100,000
such ASes on the Internet today. Each AS is called “autonomous” because each is a network administered
independently. When these networks interconnect, they constitute the public Internet as we know it. Every
Internet user is always a part of one of these Autonomous Systems.

To move information around the Internet there

needs to be “routes” through the map of
interconnected networks. The protocol that makes
the interconnections and map possible is called the
Border Gateway Protocol (BGP). BGP is the protocol
that allows computers to map a route for packets
from source to destination using the shortest and
cheapest path possible.

Each AS uses BGP to build its own map for traversing

the Internet. However, almost no BGP server has a
complete global map of all the possible paths for
packets on the Internet.

When a node sends a packet onto the Internet, it

usually sends the packet to the first router it knows
about. This is often the primary router of the ISP that
the node is connected to. That router reads the
packet and looks at the header to decide if the
packet is destined for a node that is connected to the
same network as the router. If the packet is destined
for a node on the ISP’s own network, the ISP can
deliver it on its own. If the address indicates that the
packet must be sent onward, it uses BGP to decide
which new router to send it to. Every router on the
path does this until the packets arrive at their
destination.

Figure 5.1: Routers in the Internet

When BGP servers of different ASes talk to each other, they become neighbours. As neighbours, they exchange
maps of the routes they know about and want to share. Shorter paths for the packets are preferred because that
results in less time for the packet to get to its intended destination.

Once the ASes have a map of available routes with their neighbours, they can exchange data. In some cases, the
ASes will agree to exchange traffic between the ASes for free – an arrangement called “peering.”

Often the size of the neighbours will be different. The word “transit” is used to describe connectivity to any
destination on the Internet.

© 2022 Plum Consulting 11

A Busy Person's Guide to How the Internet Works (and is paid for) 5 How Does Information Move Around the Internet?

Any organisation responsible for part of a packet’s journey from source to destination has a stake in peering and
transit. This includes access providers of any size. ISPs who provide access – using any access technology – need
a way to route the packet toward its destination. In addition, the access provider also provides a path to its own
customers. That way, packets headed to an ISP customer’s endpoint have a path for delivery. This novel
distributed way the Internet works contrasts fundamentally with the way traditional telephony networks worked
historically, linking A to B in a linear, dedicated straight line, rather than through several diverse nodes as the
Internet does.

© 2022 Plum Consulting 12

A Busy Person's Guide to How the Internet Works (and is paid for) 6 An Example of How Content is Delivered

6 An Example of How Content is Delivered

Imagine an instructional video produced in South Korea being viewed by an Internet user in South Africa. How
would the video be delivered?

1. First, the video content would be created and produced by a user or company in South Korea.

2. Once the video was prepared, it would be transmitted to the Internet via that user’s local ISP.

3. The ISP would use its transit and peering resources to send the video to the local router of the video
service.

4. Then, the video service would use its own backbone and private networks to forward the video to the
video service’s data centre for processing.

5. Once processed, the video is then transmitted to the video service’s local data centres around the world
– once again using its own backbone, subsea and private networks to distribute the newly processed
video to local servers run by the video service.

6. At this point, the video is now ready to be accessed locally by users around the world.

7. When the user in South Africa requests the video, the request is sent by that user’s ISP through the ISP’s
network to the local video server.

8. The local video server then responds to the request with a local copy of the video and sends it back to
the requesting user via the user’s ISP. (see Figure 2.2).

© 2022 Plum Consulting 13

A Busy Person's Guide to How the Internet Works (and is paid for) 7 How is Peering and Transit Paid For?

7 How is Peering and Transit Paid For?

Physical interconnection at a router is only one side of the interconnection process. There needs to be an
agreement between the different actors so they can send their data to each other.

Transit arrangements provide access to the entire Internet, while IP Peering arrangements facilitate the direct
mutual exchange of traffic between directly connected players and each party’s downstream customers. The two
types of interconnections can be both complementary and substitutable arrangements depending on the
network configuration chosen by an operator.

A key difference between transit and peering is that in the vast majority of cases, no money is exchanged
between networks that agree to peer with one another. The benefit to an operator of an AS to peer is that the
arrangement can reduce dependency on upstream transit providers – reducing that dependency has the effect
of reducing cost for that operator.

Peering enables the two networks to exchange data and benefit equally. The peering arrangement also has the
ASes advertise only their internal customer routes. Although this is a zero-cost arrangement, there are usually
costs for both parties associated with co-location, the routes to get to the co-located facility, and the required
infrastructure connections. Strategies for peering (and interconnection in general) vary from one network
operator to another. Those may be explained in a reference document, which is public most of the time and
known as the “peering policy”, although there is rarely any requirement to stick to this policy in the network’s
dealings with others. Cloudflare, for example, has an “open peering” policy and participates at nearly 150
Internet exchanges.

In practice, peering agreements are often not covered by a written contract and are established by informal
agreements between the two parties. According to Packet Clearing House (PCH), in 2016 about 99.9 per cent
of the peering agreements were done informally using a handshake. While there are cases where one party pays
for peering in the contemporary Internet, this is relatively rare.

© 2022 Plum Consulting 14

A Busy Person's Guide to How the Internet Works (and is paid for) 7 How is Peering and Transit Paid For?

Figure 7.1: Information and Intermediaries on the Internet

Source: Plum

© 2022 Plum Consulting 15

A Busy Person's Guide to How the Internet Works (and is paid for) 8 What are Internet eXchange Points and Content Delivery Networks?

8 What are Internet eXchange Points and

Content Delivery Networks?
Internet eXchange Points (IXPs) are co-located connections of many ASes for the purpose of peering. The IXP is
usually located in a data centre which provides the infrastructure to join multiple networks together at a single
location. IXPs make it easier for networks to connect to many different ASes over a single connection.

Content Delivery Networks (CDNs) have become a major part of Internet infrastructure by contributing to a
more regionalised Internet interconnect that relies less on relatively expensive international transit. The
increasing volume of content going through the global Internet in the past few years has required fresh
solutions to be developed to avoid the following issues:

● One technical: the latency of content delivered may increase, resulting in a lower quality of service to the
user; and,

● One economic: the overall cost of content delivery would increase.

In this context, CDN services have emerged to move the content “closer” to the final user for improved delivery.
A CDN is a network of servers (called edge servers) that has been deployed in different geographical locations
that cooperate in providing content/data to their users by handling traffic loads and reducing the time of
content delivery, through a process of replication. The CDN determines the shortest possible route between the
user and the CDN (located at the edge of the Internet) based on indicators like proximity, speed, or latency.

© 2022 Plum Consulting 16

A Busy Person's Guide to How the Internet Works (and is paid for) 9 What is Content on the Internet?

9 What is Content on the Internet?

The content provision environment involves a variety of applications, services and players, but essentially online
content refers to everything an Internet user can look for online: this could be cat videos on YouTube, ideas on
Pinterest, news articles on the Washington Post or John Lennon’s biography on Wikipedia. Content provision
services vary by type, size, reach and business model, but they all share one common feature: they all seek to
capture and keep the consumer’s attention, provide needed services and distribute information.

What follows is a simplified classification based on the consumer’s intention when navigating online. It is
assumed that a consumer can either be looking for entertainment, for information, for a service, or for
socialising.

Figure 9.1: Different categories of online content

Entertainment-based Information-based Socialising-based content Service-oriented content

content content

● Video streaming ● Websites ● Social media (FB, ● E-commerce

(Netflix, Disney+) (Vodafone’s Instagram, platforms
● Online gaming website) Linkedin) (Amazon, Alibaba)
● Musique streaming ● Online newspapers ● User-generated ● Apartment rental
(Spotify, Pandora) ● Knowledge sharing content (Airbnb)
platforms ● Dating platforms
(Wikipedia) (Tinder)
● E-learning
platforms
(Coursera, Udemy)
Source: Plum

Early Internet content was primarily static. A device or application requested content from a server, the server
prepared and returned a response, and the device or application presented the results. While some of this static
content remains on the Internet, much of the content landscape has changed.

One major change is the rise of mobile applications, where an end-user takes advantage of a mobile device to
gain access to services and information. In 2010, there were only 210,000 apps in the Apple App Store. Today the
number is more than 4.5 million. The Internet acts as the connectivity foundation for the vast majority of those
apps, but the tools and protocols in use are quite different from the ones used by electronic mail or other older
services.

If mobility has changed how the Internet works, so have hyper-scale platforms. Platforms such as Facebook and
Twitter also use the Internet as the fundamental connectivity tool, but the content of these services is
dynamically generated in real-time, often by end-users, and in a state of constant revision.

© 2022 Plum Consulting 17

A Busy Person's Guide to How the Internet Works (and is paid for) 10 Who Pays for Content?

10 Who Pays for Content?

Business models vary greatly from one content provider to another, but they mainly depend on how they
allocate the space between content and advertising. There are three different business models that dominate
content economics today.

● Subscription-only business models. Content Providers (CPs) that operate on a subscription-only business
model such as Netflix and Disney+ get paid by end-customers through a flat subscription fee. They
usually require the customer to sign-up for monthly, quarterly, or annual automatic payment plans and
can generate recurrent revenue from an engaged customer base. These CPs may face the potential for
higher customer churn or loss than free services. In some cases, CPs can offer access to exclusive
content for paying members, while non-paying members are only given limited access: This is referred
to as a “Freemium model.”

● Subscription-based and advertising-based models. Other CPs, such as Spotify, get paid by both
consumers and advertisers. They may choose to use an advertising network or sell ads privately through
their own negotiation. Non-paying consumers have access to the proposed content but are targeted
with advertisements, while paying customers can usually benefit from an ad-free service.

● Advertising-funded only models. Other content providers attract consumers by offering their services for
free such as YouTube or Facebook. Although there is no monetary payment involved in this case, there
is still an exchange: In exchange for the service (watching videos, communicating with friends etc.),
consumers provide their attention and personal data. The data collected enables the content providers
to attract advertisers who are willing to pay to reach their customers based on their preferences and
their online activities. This model works best when the volume of users is very large or very specialised.

© 2022 Plum Consulting 18

A Busy Person's Guide to How the Internet Works (and is paid for) 11 The Internet and Regulation

11 The Internet and Regulation

Unlike traditional telecommunications, the Internet is entirely decentralized: connecting networks together
across the globe without concern for jurisdictional boundaries. A lack of centralized control means that explicit
regulatory control by an individual government can sometimes be circumvented. If a website is removed in one
country, it can be back in operation nearly instantly in another. If a document is removed from a website, it can
be reposted on many more almost trivially.

The result is that many public policy and regulatory issues on the Internet often cannot be handled by
traditional territorial, and national institutions. The regulatory and public policy model for the Internet is usually
not top-down but characterised by transnational cooperation between many stakeholders, including protocol
engineers, ISPs, network operators, users, governments, and international organisations. National policies play
an important part in shaping the Internet, but the Internet has also evolved new institutions and governance
arrangements that respond to the unique needs of the network.

Internet governance happens at the global, regional, national, and local levels. As a basic rule, infrastructure and
the technical layer of the Internet have a global approach to governance. Protocols, cables, and routers are
maintained collaboratively by multi-stakeholder organisations that support the Internet as a cross-border and
international technical structure. In contrast, applications and content are more susceptible to national or local
governance mechanisms. As an example, regulating content that is allowed to be published or viewed online
means that Internet users are subject to their countries’ laws and regulations when going online.

There is no equivalent to the International Telecommunications Union or national regulatory authorities as there
are in traditional telecommunications, in part due to the global nature of the Internet, and the multiplicity of
private networks that comprise it.

© 2022 Plum Consulting 19

A Busy Person's Guide to How the Internet Works (and is paid for) 12 Advantages of the Internet’s Model

12 Advantages of the Internet’s Model

The simplicity of the Internet Protocol means that it can run on top of almost any physical network. This gives
the Internet the advantage of tremendous flexibility. The Internet runs over a wide variety of physical carriers
including simple copper pairs, coaxial TV networks, cellular mobile systems, satellite systems and a wide variety
of wireless networks. The advantage of the Internet model is that there can be improvements in the physical
capacity of the underlying network without the need to change anything from the Internet Protocol to the
application. This flexibility results in the Internet Protocol dominating as the preferred method to transport most
kinds of traffic, including traditional voice, video, and other data.

Simplicity alone is not the only advantage: the Internet has scaled faster and more dramatically than any other
network technology in history. Not only has the number of people and devices grown exponentially in the last
twenty years, but the amount of traffic has increased dramatically. Once again, scaling the Internet is related to
the independence of the layers that make up protocols on the Internet. Both the networks and the applications
that use them, can be changed or replaced without impacting the other.

If we use the common characteristic of the Internet as a ‘network of networks” the growth rate would be in
evidence by counting the number of distinct networks connected together. In 1986 there were 80 Autonomous
Systems (AS) attached to the Internet, by 2000 the number had increased to about 10,000 and today the
number is nearly 100,0002. This expansion combines growth in the number of users with an expanded
geographic reach. In addition, to the number of networks, the Internet has also scaled up in capacity in the
global backbone. Transoceanic cable capacity is primarily driven by growth in content and Internet applications
and content providers are investing in the supply of that part of the backbone – those companies now account
for about two-thirds of the total capacity of subsea cables.3

The Internet also has an advantage in adapting to new needs and requirements. Before the mid-1990’s,
information publication and sharing was difficult. The emergence of a new technology, the World Wide Web,
meant that nearly anyone could publish on the Internet and that publishing application was available to anyone
around the globe. The Web emerged as a dominant technology – so much so that many users think the World
Wide Web is the Internet. The Internet has the advantage of becoming the primary tool for distributing services.
Traditional services, such as health care, travel reservations, schooling, watching movies or banking have moved
from being associated with a specific location to being available anywhere. Support for diverse applications
directly results from the permissionless innovation design principle.

Recently, not only has the Internet’s scaling advantage been obvious, but so is its resilience in the face of
unexpected changes or events. The pandemic has been the most dramatic unexpected event: traffic rates in
some regions of the world increased by 50 per cent over the previous, pre-pandemic year. Working from home,
homeschooling and home entertainment all combined to create a sudden explosion of traffic. Developers of
applications worked to change their applications to work in the face of increased volume. Operators of networks
also increased network capacity in the face of the increase in traffic. Studies have shown that the deployment of
increased capacity allowed end-users to maintain download speeds without implementing service guarantees or
class of service rules. One reason for this resilience is that the Internet is decentralised, and the networks that
make up the Internet are autonomous and able to respond to local changes in demand independently of other
parts of the network. Another is that the Internet's topology contains no single, central point of control. There
are multiple routes to most content on the Internet, and if one of those routes fails, other routes can provide
needed content without other changes to the network

2
IPv6 CIDR Report, May 2022. Available at: https://www.cidr-report.org/as2.0/
3
Aqua Comms, High Tide: the state of subsea cables. 2016. Available at: https://www.datacenterdynamics.com/en/opinions/high-tide-the-state-of-
subsea-cables/

© 2022 Plum Consulting 20

A Busy Person's Guide to How the Internet Works (and is paid for) 12 Advantages of the Internet’s Model

© 2022 Plum Consulting 21

A Busy Person's Guide to How the Internet Works (and is paid for)

© 2022 Plum Consulting London LLP, all rights reserved.

This document has been commissioned by our client and has been compiled solely for their specific
requirements and based on the information they have supplied. We accept no liability whatsoever to any party
other than our commissioning client; no such third party may place any reliance on the content of this
document; and any use it may make of the same is entirely at its own risk.

© 2022 Plum Consulting 22