How Does the Internet Work and Who Keeps it Working?

Musab Isah, PhD


To understand how the internet works, we need to start with the history of how the internet worked in the past. The internet came from a United States Department of Defense’s project established by the Advanced Research Projects Agency (ARPA) called ARPA Network or ARPANET for short. ARPANET started by connecting 4 computers located in four different places in the US, the University of California – Los Angeles (UCLA), Stanford Research Institute (SRI), University of California – Santa Barbara (UCSB), and the University of Utah.

The first successful computer to computer (or host to host, in networking parlance) connection was achieved on 29 October 1969 and the four hosts were fully connected to each other by 5 December 1969. It was from this four-host network that the number of devices on the internet grew to 10 billion today; and we are still counting. These four hosts were directly connected to each other using some direct wiring with cables allowing the institutions involved to access files located on the different hosts and to transmit information between them. Note that every resource or program we access on the internet today is stored in, or as, a file on another computer somewhere around the world.

Figure 1: ARPANET in 1974 showing connectivity within the US.
Source: Wikipedia, https://en.wikipedia.org/wiki/ARPANET

This new method of sharing resources was first utilised within institutions in the US, forming local networks between different computers in the institutes. This local network is then connected to other local networks in the US (and 2 locations in Europe by 1974) forming a network of autonomous networks or internetwork or internet for short (see Figure 1). A suite of protocols named Transport Control Protocol/internet Protocol (TCP/IP) was developed to serve as the computer programmes that allow devices to send packets (data in transit) to each other over the internet. TCP/IP suites are still the main protocols powering the internet today. IP addresses are used by the sending node to declare where the packets should be delivered to on the internet and TCP is used to identify the application – such as an email server/client, a computer browser, etc. – which should handle the packets at both the sender and the receiver.

Commercial use of the internet began when phone companies connected to ARPANET and allowed their subscribers to access resources and transmit information on the internet using computers at home. The companies that provide access to the internet for their subscribers are today known as internet Service Providers or ISPs for short. To simplify locating of resources and to manage the increasing number of users on the internet, Domain Name System (DNS) was created in 1983 to allow the use of domain names instead of IP addresses. For example, https://afrisig.org/ is easier to remember for humans than the IP address 193.37.35.226 of the same domain. I will come back to DNS later.

Figure 2: Internet structure – a “network of networks”.
Source: Slides of Computer Networking: A Top-Down Approach, 8th edition by Jim Kurose, Keith Ross, Pearson, 2020

For ISP A clients to access resources on ISP B or vice versa, the two providers needed to run their cables to each other’s network. As the number of ISPs grows, commensurate with the growth of users, it was no more possible for all ISPs to connect to each other, so Internet eXchange Points (IXPs) were established to serve as meeting points for the different networks operating within a given geographic area (see Figure 2). Most ISPs plug their cables at the IXPs, thereby sharing what they have with many other ISPs using only one connection point. IXPs are also vital in avoiding the trombone effect – a situation whereby a message from one ISP destined to another local ISP network had to travel to a different geographical location before coming back to its destination just because the two ISPs were not directly connected.


Some small ISPs only need to connect to a big ISP to access the world as the latter ISPs are connected to many other providers. The growth in Internet usage also saw the Universities and research institutes also forming their networks (e.g. Internet2 in the US and JANET in the UK) and connecting to the ISPs directly or via the IXPs. For a gateway router (the high-capacity computing device that connects one autonomous network to another or to an IXPs) to know whether incoming packets are destined to devices within its network or otherwise, routing protocols were introduced such as Open Shortest Path First (OSPF) for routing within an autonomous network and Border Gateway Protocol (BGP) for routing between independent networks.


The need to share and access resources to and from different parts of the world means we need to run cables underground and underwater in other for devices in different parts of the world to join this interconnected network of computers. Some big organisations – such as big ISPs and university networks – built these land and underwater cables, and other ISPs pay the cable owners to have their network traffic transmitted around the globe. The connection between autonomous networks across the world is referred to as the core or backbone network of the internet while the part of the network that allows users to have access to the internet is referred to as Access Network. Most ISP to ISP/IXP connections today are achieved using fibre optic and Ethernet technologies connecting routers and switches, the two main networking devices. Also connected to the ISPs/IXPs are content delivery networks (CDNs), a group of geographically distributed servers meant to bring web content closer to users and to improve availability and performance.


Access Networks and Data/Subscription
Access networks come in different forms including cellular networks – 2G, 3G, 4G, and, recently, 5G – from mobile telecommunication companies (or telcos for short), Wi-FI networks, fibre to the home (FTTH), Ethernet, and Digital Subscriber Line (DSL). Since telco networks are the most common means of access to the internet especially in the African region, I will focus on explaining how the internet works from that perspective. I will also touch on the data or subscription that we pay for in order to obtain internet services using an operator’s network.


A telco is an ISP that provides telephony and data communications services. A company usually installs infrastructures including base stations (also called radio mast or telecommunication/cell tower), fibre optic cables, and several networking devices at the core of its network to enable calls and internet services for its customers. A telco also connects its network to the internet through other ISPs and/or IXPs. So, a user’s internet traffic is sent wirelessly to the antennas mounted on the base station. The base station is connected, using the laid fibre optic cables, to the core network of the telco, and the telco’s core network is connected to the internet via the cables between the telco’s core network and the ISPs/IXPs.


When you open the Facebook app to access your profile, your telco needs to send the request to the nearest Facebook server for your profile to be retrieved. Your internet traffic will likely travel through some intermediary network(s) before it gets to the Facebook server and your telco must have initially paid for bandwidth (the right of passage through the intermediary networks) to allow for the traffic to flow through. And that is why the telco charges you for data because it also pays other networks to send your data over the internet.


This brings us to the question of what is data plan or subscription that allows one access to the internet is? A data plan is a subscription you pay for with your telco to enable you to access the internet on your phone (or on other wireless devices). It is similar to minutes of calls available on your phone but the consumable in this case is the data. Now, what is data? Data, in this context, is anything that you could only access if you have the internet on your phone: emails, WhatsApp messages/calls, videos, Facebook, newspapers, etc. Your data plan expires after a period because telco’s pay for bandwidth with other ISPs on a termly basis and they also have to pay again upon expiry of the term whether they use all the bandwidth or not.


When you subscribe to 1GB of data on your phone, it simply means that you could access data (files) on the internet (and/or send files to the internet) to the size of 1.07 billion bytes or 1.05 Million kilobytes or 1,024 megabytes before your allowance is exhausted. The byte is the measure of file sizes in computing and everything you access on the internet is stored somewhere as a file. To put things in perspective, a low-quality video (240p) will use around 1.6MB per minute, but a high-definition HD (1080p) video will use as much as 12MB per minute. Whatsapp messages, Facebook textual posts/comments, emails, and any other text-based data download/upload will usually amount to very little usage of your available data. Every post, comment, email sent, etc is an upload; reading email, posts, comments, etc, is download.


It is pertinent to point out the difference between KB (MB/GB) and Kb (Mb/Gb – with small ‘b’). The first units, with ‘B’, as explained earlier, show the size of files (amount of data) while the second unit, often written as kbps (mbps/gbps), meaning kilobit per second, is the speed of your telco’s cellular network. For instance, if the speed is given as 1mbps (you may find out by googling ‘speed of my network’ and follow the links; the speed in your operator’s terms and conditions is the maximum that could be attained in an ideal condition), you can access a youtube video at (128 kBps), which translates into videos in 360p/480p quality. In simple terms, the size of your data plan (MB/GB) does not determine the quality of internet you get from the operators. Always check the speed.


Lastly on telcos, the G used to describe a cellular technology – 3G, 4G, and 5G stands for 3rd, 4th, and 5th Generation respectively. Every few years, telecom engineers and researchers come up with improved technologies to increase internet speed – especially – as well as other services on mobile networks. For e.g., 1G (1st Generation Cellular) offers 2.4 kbps (notice the small ‘b’), 2G offers 64 kbps, 3G offers 144 kbps–2 mbps, 4G offers 100 Mbps–1 Gbps, and the max speed of 5G is aimed at being as fast as 35.46 Gbps, which is over 35 times faster than 4G.

A Day in the Life of a Web Request
Let us consider a situation where you switch on your phone in order to access the URL, https://afrisig.org/. Your device will go through the following processes before the page is displayed on your browser.
1. Your phone will authenticate with your telco using the nearest base station to you and an IP address will be allocated to your phone by a DHCP server set by the telco. Other information, such as the IP address of default gateway (the router through which all your device traffic will pass) as well as the DNS server will also be provided to your device.
2. After you type in the URL above, your phone needs to resolve the domain name with the IP address since it can only send the request for the page using an IP address. The IP address tells your phone of the web server that is hosting the web page you seek. So, a DNS query for the URL is sent to the DNS server using the DNS server’s IP address provided in step 1.
3. The DNS server responds with the IP address of the requested domain/web server or, in the case that it does not have the records of the domain, it provides the IP address of another DNS server who knows. The IP address 193.37.35.226 is returned to the phone.
4. Once the IP address is received, the device sends a request for the page using HTTPS (or HTTP) to the web server. HTTP (the additional ‘s’ is for secure version) is the protocol that enables a browser to request for web pages and enables a web server to understand the requests and respond to them.
5. The web server responds with the page data, which is rendered for view by your phone browser.

Who Keeps the internet Running
As we might have learned from our discussion so far, the internet is a distributed network joined voluntarily by different autonomous systems. So, if each autonomous system does what it is supposed to do – use the right protocols, do the right configurations on network devices, use the right standards on technologies in use, use the right IP addresses, etc – then the internet will continue to run just fine. However, determining the right thing to do, which is essentially governing the internet, is a process that is conducted by multi-stakeholders including governments, civil society, researchers, academics, private sector, national and international organisations, among others. Examples of these internet stakeholders include routing information registries (RIR), internet corporation for assigned names and numbers (ICANN), internet governance forum (IGF), internet engineering taskforce (IETF), etc. These bodies perform many functions including advising, educating, designing policies, conducting research, coming up with standards, discuss operations and regulations, among many other functions.

Dr Musab Isah

https://orcid.org/0000-0001-7108-7631

I am an Assistant Professor of Computer Science and Engineering at the University of Hafr Al-Batin, Saudi Arabia. I teach courses in Data Science and Computer Networks.

Prior to my current role, I worked as a Research Engineer and Data Scientist at AFRINIC. I planned and delivered research projects which were focussed on different aspects of Africa’s Internet from routing protocols to DNS to IPv6 adoption. I also travelled across Africa to deliver workshops and have participated in Internet fora and conferences with the drive to have a common understanding of technologies and policies that are vital in ensuring that the vast majority of unconnected Africans are provided with cheap access to the Internet.

I was also a Research Associate at the University of Cambridge and worked on the aRchitecture for an Internet For Everybody or simply the RIFE project. RIFE was aimed at providing affordable Internet access to those who could not afford it by solving the technological challenge to increase the efficiency of the underlying transport networks and the involved architectures and protocols. I also worked in a project to develop a new mobile web framework designed to revolutionise web access through significant reduction of the sizes of web pages in order to speed up web access especially for users in underserved areas. I am proud to say that the work we did contributed to the formation of the company GAIUS Networks.

My research interests are in Internet Data Science, Internet Performance measurement, mobility on the Internet, mobile web performance, mobile network architectures, Internet provision in underserved areas, and the general interaction between the different players in the Internet Ecosystem.


Publications


Talks

  • Internet Measurement for Regulators, Africa Internet Summit, June 2019, Kampala, Uganda.
  • State of Internet Measurement in Africa – A Survey, Africa Internet Summit, June 2019, Kampala, Uganda.
  • Libremesh on Alpha Router Platform, Ammbr Research Lab & Product Foundary Meeting, 10th April 2018, Universitat Politècnica de Catalunya (UPC), Barcelona, Spain
  • Emulation Experiments Platform for Mesh Networks – An MLC Introduction, 15th March 2018, University of Cape Town, South Africa
  • Mesh Networking Research @AMMBR Research Lab, AMMBRTech Stakeholders Meeting, 20th January 2018, Brussels, Belgium
  • Towards Zero-Packet-Loss with LISP-MN, IEEE International Conference on Computing, Networking and Communications, ICNC 2017, Silicon Valley, California
  • Inter-Domain Mobility with LISP-MN – A Performance Comparison with MIPv6, IFIP WMNC 2015, Munich, Germany
  • An ILNP-Based Mobility Solution for Future Heterogeneous Wireless Networks, PGNet 2013, Liverpool, UK

Education