A single, isolated computer can only ever work with what is physically stored on it. The moment you connect it to even one other machine, it gains access to resources it does not own itself, and that trade-off, giving up a little independence for a lot of shared capability, is the entire reason networks exist.

A network is two or more computers connected together so they can communicate and share resources. Networking devices such as switches, routers and wireless access points exist purely to make that sharing possible, physically carrying the signals between machines and directing them to the right destination.

Networks are classified first by size, because the physical distance data has to travel completely changes what hardware and infrastructure you need to move it.

Once you have a network, someone has to decide how the computers on it relate to each other: does one machine take charge, or does every machine treat every other machine as an equal? That decision is the network model, and it shapes everything from cost to security to how the network behaves when something fails.

Some networks combine ideas from both: a subnetwork model splits a larger network into smaller, more manageable sections (subnets), each of which might itself use a client-server or peer-to-peer arrangement internally, connected together through routers.

In a client-server network, individual client machines can be built in two very different ways, depending on how much of the actual work they are trusted, or equipped, to do themselves.

A topology is the pattern in which devices on a network are connected to each other. It is not just a diagram, the topology a network uses directly determines how data physically travels from one host to another, and what happens to the rest of the network when one link or device fails.

Cloud computing provides on-demand access to computing resources, servers, storage, databases and software, delivered over the internet rather than run on hardware you own, typically billed on a pay-per-use basis. Instead of buying and maintaining your own server room, you rent capacity from a provider like AWS, Microsoft Azure or Google Cloud.

Public vs private cloud

A public cloud is shared infrastructure, owned and operated by a provider, that many different customers rent capacity on simultaneously, your data lives alongside other organisations' data, logically separated but on shared physical hardware. A private cloud is infrastructure owned and operated by, or exclusively for, a single organisation, keeping full control over the hardware while still gaining cloud-style flexibility.

Every network link needs a physical medium to actually carry its signal, and the choice of medium is a genuine engineering trade-off between speed, distance, cost, and how easily the signal can be disrupted.

Wired vs wireless: the underlying trade-off

None of the topologies in the previous section work without physical hardware to actually create and manage the connections. Each device below has a distinct, non-overlapping job, and exam questions frequently test whether you can tell them apart.

The router: connecting networks together

Every device above operates within a single LAN. A router is different: its job is to connect separate networks to each other, most commonly connecting your home or school LAN to your ISP's network, and ultimately to the internet. A router examines the destination IP address of each packet it receives and decides which network to forward it towards next, essentially acting as a junction and signpost between networks, rather than just a distribution point within one.

Ethernet is the standard technology for wired LANs. On a shared medium, nothing physically stops two devices from trying to transmit at exactly the same moment, and when that happens, their signals corrupt each other in a collision. Ethernet networks manage this with a protocol called CSMA/CD, Carrier Sense Multiple Access with Collision Detection.

How CSMA/CD works, step by step

1. Carrier sense: before transmitting, a device listens to the shared medium to check whether it is currently idle.
2. Multiple access: the medium is shared, so several devices are able to attempt transmission independently.
3. Transmit: if the channel appears idle, the device begins sending its frame.
4. Collision detection: if two devices happen to transmit at the same moment, their signals collide and corrupt each other, both devices detect this.
5. Jam signal: both devices immediately broadcast a short jam signal, ensuring every device on the network knows a collision occurred and discards the corrupted data.
6. Random backoff: each device involved waits a random amount of time before attempting to retransmit, making it unlikely they will collide again on the retry.

Bit streaming is the continuous transfer of data, most commonly audio or video, as a steady stream of bits that can start being played back before the whole file has finished downloading. This is what lets you start watching a video within a second or two, instead of waiting for a multi-gigabyte file to download completely first.

Bit rate, broadband speed, and buffering

A video file has a bit rate, the number of bits needed per second to play it back smoothly, and a higher bit rate generally means better picture and sound quality. For playback to be smooth, your broadband connection's actual download speed must keep up with, or exceed, that bit rate. When the connection is slower than the bit rate, the player's buffer empties faster than it can refill, and playback has to pause, buffering, until enough data has arrived again.

People use "the internet" and "the web" interchangeably in everyday speech, but the syllabus specifically expects you to explain why they are not the same thing. Confusing them is one of the most common ways to lose a mark on this topic, so it is worth being precise.

For a packet to reach the right computer out of billions connected to the internet, every device needs a unique address. That is the job of an IP address, the internet's equivalent of a postal address.

How an IP address gets associated with a device

A device does not usually have its IP address permanently baked in. When a device joins a network, it is typically assigned an IP address automatically by a server running DHCP (Dynamic Host Configuration Protocol), which hands out an available address from its pool and keeps track of which device is using which address for as long as that device stays connected.

Subnetting

Subnetting divides one large network into several smaller sub-networks (subnets), using a subnet mask to split every IP address into two parts: a network portion, identifying which subnet a device belongs to, and a host portion, identifying that specific device within its subnet. Wherever the subnet mask has a 1 bit, that position belongs to the network portion, wherever it has a 0, that position belongs to the host portion.

Typing painlessprogramming.com into a browser is far easier than remembering an IP address, but underneath, computers still only route traffic using IP addresses. Two systems bridge that gap: the URL, which tells the browser what to ask for, and DNS, which translates the human-readable name into the numeric address a computer can actually use.

URL structure

A URL (Uniform Resource Locator) identifies exactly where a specific resource lives on the web.

How DNS resolves a domain name to an IP address

The Domain Name Service (DNS) works like a phonebook for the internet, it looks up the domain name you typed and returns the IP address of the server that hosts it, so your browser knows exactly where to send its request.

Try each of these before reading the model answer. They are written in the same style as Cambridge 9618 Paper 1 questions.