Internet for beginners - So what is the Internet?

You often hear "internet" as well as "web" and may think they refer to the same thing as they get used interchangeably. But like a lot of every-day-assumptions, it's a mistake. It doesn't help when false analogies and buzzwords, along with a lot of misattributed jargons, just add to the confusion. The internet just seems to be this ethereal thing that's "there", that one has a connection to or is present on their device like an "app", yet also all these websites and maybe others like e-mail. It's even often forgotten or not even realised that the internet is really just a telecommunications system, which means it's for remote communication but for computers as opposed for people. Sometimes even "the internet" is a phrase for "people", "social media" or "online chatter". Yet it is not. It never was made for social media, it was originally only made to help computers share data, for military and academic purposes. This article defines "internet" in a more technically concise way, so read on.

The technical explanation:

The internet, short for "interconnected networks", is a global-spanning packet-switching computer network or rather a "network of networks". It's a mostly non-centralised network of interconnected users (computers) that facilitate data exchange regardless of the current physical and local network one is connected to. It's a digital information medium comprised of various electrical and physical technologies in which nodes (computers or network equipment such as routers) adhere to a common protocol (set of communication rules) known as the Internet Protocol (IP). Currently, as of 2022, there are two versions of IP in use: IPv4 and IPv6, which are not directly cross-compatible software-wise although they can run over the same hardware. All data that traverses the internet is arranged in chunks (packets) prepended by a "header", which mostly dictate where to and from the packet or message is intended, as well as other data for routing and handling.

The internet was built upon other packet-switching networks, which included the ARPANET which was the first, in order to unify them into one common network. Each end node or client could be assigned a unique number (IP address) that allowed any to send messages (or packets) to others providing they have sufficient internet connectivity. The nature of such a system however cannot guarantee speed/flow rate (bandwidth) nor ensure reliability like a telephone call, but does allow many-to-many communications at the same time that is, in principle, always available - i.e. "constant connection". Around the conception of IP version 4, the transport layer protocol (which is encapsulated by IP - meaning it sits within) known as Transmission Control Protocol (TCP) was developed to supplement the IP, bringing it methods and standards for reliability and traffic (data flow) management, such as detecting packet loss and putting them in order before the system delivered the complete data to the end user or application (program). This became widely known as the "TCP/IP suite", however the internet supports other transport-layer protocols such as User Datagram Protocol (UDP), which gives a smaller and simpler header at the cost of reliability and flow control. The latter allows one (software engineers) to implement their own session-tracking and traffic-problem managements without having to be forced to use so many included with TCP.

IP suite OSI model

The protocol stack to allows the internet services to work depends on several aspects bridging the physical and software. The highest level protocol - the application layer- can run without regard of hardware protocols (like Wi-Fi or ethernet link layers or electrical signalling) and vice-versa. However, applications design do need to deal with IP versions as IP addresses are vital for communication and transport layers only manage data delivery (or traffic) to the designated client or peer program. Below shows the TCP/IP suite OSI model:

Layer no. Name Description
4 Application layer Defines methods and language that can be understood between client and server or peer-to-peer. Provides presentable data for the end user's program. Examples include HTTP(S), DNS, SMTP, IMAP, POP, DHCP and FTP. Applications can use their own exclusive protocol, as often is the case with online games.
3 Transport layer Distinguishes and moderates individual connections or sessions between end user's programs and their peer and may arrange and manage datagrams traffic in lieu of the application/user. Examples include TCP and UDP. ICMP (Internet Control Message Protocol) is not a true transport protocol, at least in practice, and is typically just used for diagnostics and some configuration.
2 Internet layer Seldom also called the "network layer", however the Internet Protocols (including ICMP and ICMPv6 which are virtually transport protocols) take in account for cross-network communication. Allows remote clients (application connections) to engage with others. IPv4 and IPv6 are by far the most common, whereas most if not all others are IP based.
1 Link layer Typically deemed the lowest level, link layer protocols establishes methods and protocols to allow devices or NIC's (network interfaces) to communicate and exchange packets with each other, limited to either NIC-to-NIC over a cable or between wireless transceivers, or within a local area network with a common network switch or hub. Datagrams/packets/messages here are often called frames and may have "footers" as well as headers. Examples include MAC (used in ethernet and Wi-Fi) and PPP (similar to ethernet, but used for DSL and dial-up broadband connections).
0 Hardware layer Not officially part of the model, physical (hardware) devices need common protocols or standards right down the signalling in order to communicate effectively, especially as different vendors or manufactures may be involved between devices and components. This factors in electronic and circuit design and transcribing digital data for the computerised system - and supported link layer protocol.

"Layer 5" is often jokingly referred to the "meat layer" or "human user layer", some arguing that this is the primary cause of problems which should be dealt with first. Indeed, human error can be the most common cause affecting not just the user application but everything else, by usage or in design and development!

The physical:

Physically the internet is an extensive network, that's highly scalable, mostly consisting of fibre-optic cables as well as networking devices, electrical cables and, at least for end users or local area networks, wireless or radio equipment. The IP is largely independent of such equipment, as it's up to said equipment and software to operate within their scopes. For example, fibre-optic networks have to encode and decode digital information (and perhaps analogue too) within beams of light or infrared, while network interface controllers transmit and receive digital data and interface between the transceivers and the computer system, which has to be designed and prepared to understand this data and handle it correctly, until a designated program deals with it. The Internet Protocol does not concern the ethernet or fibre link or radio transceivers or vice-versa. It's basically just a software thing, so only network layer protocols (like ethernet frames), transport layers and application layer needs to consider it and indeed the IP also notes what transport layer protocol it carries too. This means that messages transmitted over one link that has to be retransmitted over another (to another device) must be "routed" as per the IP settings and rules. Nonetheless, networks still need real connections that bridge them from one to the next.

Physically the internet is an extensive network, that's highly scalable, mostly consisting of fibre-optic cables as well as networking devices, electrical cables and, at least for end users or local area networks, wireless or radio equipment. The IP is largely independent of such equipment, as it's up to said equipment and software to operate within their scopes. For example, fibre-optic networks have to encode and decode digital information (and perhaps analogue too) within beams of light or infrared, while network interface controllers transmit and receive digital data and interface between the transceivers and the computer system, which has to be designed and prepared to understand this data and handle it correctly, until a designated program deals with it. The Internet Protocol does not concern the ethernet or fibre link or radio transceivers or vice-versa. It's basically just a software thing, so only network layer protocols (like ethernet frames), transport layers and application layer (like website's HTTP and e-mail's SMTP) needs to consider it and indeed the IP also notes what transport layer protocol it carries too. This means that messages transmitted over one link that has to be retransmitted over another (to another device) must be "routed" as per the IP settings and rules. Nonetheless, networks still need real connections or links that bridge them from one to the next.

The analogies:

The most accurate and simple analogy is in fact the internet is like the postal system, which delivers envelopes containing messages. The IP protocol is mostly the standard format of the envelope, containing a destination address (and source address) as well some other pieces of information. The letter inside (payload) may also have a header containing more info about it and for the specific end-user (transport layer header), while the payload of that contains the conveyed information for the recipient (application). No receipt for delivery is necessary, although protocols like TCP always attempts to acknowledge receptions. Similar to postal mail, IP packets and what they convey are delivered from one point to another until it reaches its intended destination but (usually) cannot be tracked or managed from either ends - this is "best effort delivery". Fortunately though, the internet usually allows speedy retransmissions with minimal cost. Unfortunately though, messages are intercepted by "routers" and can, in exceptional cases, can be read "on the fly" by unwanted 3rd parties (hackers or eavesdroppers) even without end users knowing thus scrambling methods (encryption) by end users and between network nodes (e.g. between Wireless/Wi-Fi Access Point and wireless client) should be employed.

Although some internet connections share some electrical infrastructure with the telephony network (e.g. xDSL broadband and dial-up), they only draw similarities in that they:

  • Use numerical identifiers (phone numbers and IP addresses)
  • Allow bi-directional end-to-end communications
  • Information may transit multiple networks

Internet Exchange Points (IXP) are locations or buildings of which multiple network owners' (AS) equipment (routers or switches or shared line cards) can be found, in order to exchange traffic. This is like a telephone exchange, international mail sorting office or trading market. However, network operators and data centres may have network equipment sited in other, more niche, places too, listed as "Point of Presence".

Informal and inaccurate synonyms:

Interweb, the net, the web, cyberspace, Wi-Fi, the cloud, broadband, 2/3/4/5G (or other cellular/mobile connections), big tech, social media, computer hivemind, series of tubes, cable(s).