HookThe day the internet's address book went down
On 21 October 2016 large parts of the American internet went dark. Twitter, Netflix, Reddit, Spotify and the PlayStation Network all stopped loading - not because their own servers had failed, but because a botnet of hijacked webcams and home routers, called Mirai, flooded a company named Dyn with junk traffic. Dyn ran DNS, the internet's address book, and with the address book unreachable, browsers could no longer turn a name like twitter.com into the numeric address of the machine behind it. The websites were perfectly healthy; the problem was that nobody could find them.
That outage is a neat way into section 1.3, because it touches almost every idea in it: what a network actually is, how the internet is stitched together from smaller networks, the DNS lookup that quietly happens before every page load, and the hardware, addresses and agreed rules - protocols - that carry your request there and back. OCR asks you to explain how devices are joined into a network and how they talk once they are: LAN versus WAN, client-server versus peer-to-peer, star and mesh layouts, wired against wireless, the difference between an IP and a MAC address, the handful of protocols you must name on sight, and why the whole job is split into layers.
ModelLAN, WAN and the network of networks
A network is two or more devices connected so they can share data and resources. Scale gives you the first distinction. A LAN (local area network) covers a single site - a home, a school, an office - and the organisation owns all the cabling and hardware itself. A WAN (wide area network) connects devices across large distances, so it must use infrastructure the organisation does not own, such as fibre, telephone lines and satellite links leased from telecoms companies. The internet is the largest WAN of all: a worldwide collection of interconnected networks.
Three services turn that raw connectivity into something usable. DNS (the domain name system) is the address book that translates a human-friendly name like vofti.com into the numeric IP address of the server hosting it. Hosting is renting space on a powerful, always-on web server so that your site or files stay available to clients - the browsers and apps that request them - at any hour. The cloud is simply storage and processing carried out on those remote servers over the internet rather than on your own machine, which is why your photos survive a lost phone and a document opens on any device you log in from.
CaseClient-server versus peer-to-peer
Networks organise their devices in one of two ways, and OCR wants you to match each to a scenario. In a client-server network one or more powerful servers hold the files, manage security and provide services, while the clients - the ordinary computers - request them. A school network is client-server: log in at any machine and a central server checks your password and serves your files. Management, backups and security are all centralised, which is efficient and easy to control - but the server is expensive, and if it fails everyone is stuck at once.
In a peer-to-peer (P2P) network every device is an equal peer that both provides and requests resources, with no central server. Two laptops sharing files directly, or the BitTorrent system where each downloader also uploads to others, are peer-to-peer. It is cheap and simple, with no server to buy and no single point of failure - but the files are scattered across machines, so backups and security become each device's own problem, and performance drops as peers are switched off. When a question hands you a situation, read for the deciding clue: a need for central control, security and reliable backups points to client-server; low cost and a small number of equal machines points to peer-to-peer.
MechanismThe hardware, the topology and what slows a network down
To join stand-alone computers into a LAN you need specific hardware. Every device needs a network interface controller (NIC) to connect to the network at all. A switch links devices within one LAN and is clever about it: it learns which device is on which port and sends each message only to its intended recipient. A router connects different networks together and directs data between them - the box at home is really a router, a switch and a wireless access point built into one unit. A wireless access point (WAP) lets devices join without cables. The transmission media - copper Ethernet cable, fibre-optic cable or radio waves - carry the actual signal.
Two topologies are named on the spec. In a star topology every device connects to a central switch; it is fast, and one broken cable takes out only the device on it, but the whole network leans on that central switch and it uses plenty of cabling. In a mesh topology devices connect to many others, so data has several possible routes and the network keeps running even if one link fails - very resilient, but expensive and complex to wire fully. Performance itself depends on bandwidth (how much data the link can carry), the number of users sharing it at once, and the transmission medium: fibre beats copper, and both beat congested Wi-Fi.
MechanismWired, wireless and how devices are addressed
Devices connect either wired, over Ethernet cable - faster, more reliable and more secure, but you are tethered to the socket - or wireless. Wi-Fi gives convenient medium-range access over radio; Bluetooth handles short-range links between a few devices, like headphones to a phone. Because wireless signals travel through open air where anyone nearby could intercept them, encryption scrambles the data with a key so that an eavesdropper captures only unreadable ciphertext.
Every device carries two kinds of address, and confusing them is a classic dropped mark. A MAC address is a permanent 48-bit identifier burned into the NIC when it is made, written as six pairs of hex digits (for example 2C:54:91:88:C9:E3); it never changes and identifies the physical device on the local network. An IP address is assigned by the network and can change; IPv4 uses 32 bits written as four numbers from 0 to 255 (for example 192.168.1.15) and identifies where a device sits on the wider internet. Standards - agreed published rules such as the Ethernet and Wi-Fi specifications - are what let hardware from rival manufacturers work together at all.
An IPv4 address is four octets, each a denary number from 0 to 255, and each octet is just one byte of binary. Take 192.168.1.15 and convert each part: 192 = 128 + 64 = 1100 0000; 168 = 128 + 32 + 8 = 1010 1000; 1 = 0000 0001; 15 = 8 + 4 + 2 + 1 = 0000 1111. So the full 32-bit address is 11000000.10101000.00000001.00001111. Because each octet is exactly one byte, the largest value any part can hold is 1111 1111 = 255 - which is precisely why you never see an address part of 256 or more. A MAC address does the same job of identifying a device but uses 48 bits shown in hexadecimal, giving vastly more unique values so that no two NICs ever share one.
ModelProtocols and the point of layers
A protocol is an agreed set of rules for how devices communicate; without shared rules two machines just exchange noise. You need to recognise the common ones by their job. TCP/IP is the master pair that underlies the internet: IP routes packets to the correct address, while TCP splits data into packets, numbers them, reassembles them in order at the far end and resends any that go missing. HTTP fetches web pages; HTTPS is the same but encrypted, so passwords and card numbers stay safe. FTP transfers files between computers. Email uses three: SMTP sends mail out, while POP and IMAP retrieve it - POP downloads messages and typically removes them from the server, whereas IMAP leaves them on the server so they stay in sync across all your devices.
Those protocols are organised into layers: the job of communicating is divided into self-contained levels - roughly application, transport, internet and network access - each handling one part and passing its result to the next. Layering matters because a change in one layer, such as a new Wi-Fi standard, does not force a rewrite of the others; different developers can work on different layers at once; and every manufacturer follows the same divisions, so equipment from different makers interoperates. It is the same 'break a hard job into simple, independent parts' idea that runs right through computer science.
VocabularyKey terms the mark scheme pays for
TrapsMisconceptions that cost marks
ExamWhat examiners want
Paper 1 (J277/01) tests networks with a mix of short recall and extended, levelled answers. For AO1 recall, learn the protocols as a lookup table you can reproduce under pressure: match the acronym to its one job (SMTP sends mail, IMAP and POP retrieve it, HTTPS is encrypted HTTP, FTP moves files). For addressing questions, be precise that a MAC address is permanent and physical while an IP address is assigned and can change, and if asked, show an octet converting to its 8-bit binary with the place values written out.
The higher marks come from comparison and justification. A 6- or 8-mark levelled question - client-server versus peer-to-peer, or star versus mesh, or wired versus wireless - is marked on developed points, not lists. Take each factor (cost, security, reliability, performance, resilience), say how each option handles it, and then tie your recommendation back to the exact scenario, because the top band rewards a judgement, not two lists sitting side by side. When you explain DNS or the cloud, describe the process step by step (name typed, DNS returns the IP, browser requests the page) rather than defining the term in the abstract, and use the correct hardware names - NIC, switch, router, WAP - throughout.