Lesson 4: Computer Networks, Cloud Computing and Prompt Engineering

Definitions

• An information and communications network is a web-like network that uses communication lines to carry information.
• A computer network is a communication system that links information and communication devices — computers and smart phones — so that they exchange data with each other.

Types

• LAN (Local Area Network): links devices inside a limited area such as a school or a home.
• Wired LAN: data is exchanged through direct connection by cables.
• Wireless LAN: data is exchanged by radio waves instead of cables.
• WAN (Wide Area Network): links distant areas and covers a wider range than a LAN.
• The Internet: a world network that connects LANs and WANs to one another.

Connection devices

• ISP (Internet Service Provider): a company that makes Internet access available; a contract with one is needed to get the service.
• Router: a device that passes data on and re-directs it between different networks.
• Hub: a central connection point to which several cables are attached; a switch does the same job but forwards each frame only to its destination port.
• Modem: converts digital data to a signal the line can carry and back again; the NIC (network interface card) is the card that puts a device on the network, and an access point serves the wireless side of it.

• Client–server system: a server is a computer that provides various services to others, and the clients are the computers that request those services. Resources and permissions sit centrally, so management and security are stronger.
• Peer-to-peer system: every computer both supplies and consumes resources directly, without a dedicated server; cheap and simple for a small group, but weaker in security and control.

Common server types

• Web server hosts sites and answers browser requests. Mail server sends and stores messages. File server holds shared files. Database server serves queries. Printer server manages one or more printers shared over the network.

• Circuit switching: a dedicated connection path is established between the sender and the receiver before transmission starts, and the information is exchanged over it — as in a traditional telephone call. The line is reserved even when nothing is being sent.
• Packet switching: the data is split into packets, each travelling independently, possibly by different routes, and reassembled at the destination. The link is shared and no path is wasted.
• Why packets: better use of the lines, resilience (if one route fails the rest still deliver), and retransmission only of the missing packets.
• The price of packets: packets may arrive out of order, so each carries a header with sequence and address information so the receiver can reorder them.

The IP address

• An IP address is a unique identifying number given to every device connected to the Internet; it is the device's "address" on the network.
• IPv4 is divided into four blocks of 8 bits each, separated by a dot and written in decimal from 0 to 255 — for example 192.168.0.1.
• Public addresses are reachable on the Internet and are assigned by the authority; private addresses are used inside a local network and are not routable on the Internet.
• With the rapid spread of the Internet, new IPv4 addresses are almost exhausted — the solution is IPv6, whose 128-bit addresses give an enormous space, together with address-sharing techniques.

Domain names and DNS

• A domain name is the human-readable substitute for an IP address, made of parts read right to left: the top-level (organisation or country), then the second-level name, then the host label such as "www".
• Examples of generic top levels: .com commercial, .edu education, .gov government, .org organisation; country codes: .eg Egypt, .jp Japan, .kr South Korea, .cn China.
• The DNS (Domain Name System) is the server that translates a domain name into its IP address, and can answer the reverse query.

• A protocol is an agreed set of rules that lets devices understand each other; without it two machines cannot communicate.
• The TCP/IP suite works in four layers: the application layer (the service the user sees, such as HTTP, SMTP, FTP), the transport layer (TCP and UDP: segmentation, ports, reliability), the internet layer (IP: addressing and routing) and the network-access layer (the physical link and its hardware).

Steps of transmitting data over the Internet

• Split the data to be sent into packets.
• Attach the sender's and the receiver's IP addresses in the header of each packet.
• Choose the best route for each packet and deliver it to its destination — this is routing.
• At the receiving end, if packets are lost their retransmission is requested, then the packets are reordered until the data is complete.

Parts of a URL

• The World Wide Web is a system that links an countless number of Internet contents that the user browses as pages.
• A URL is read as: protocol, then host, then path — in https://www.mext.go.jp/index.html: the protocol https, the host www.mext.go.jp (with the country code jp = Japan), and the requested file /index.html.

Steps when a page is opened

• The browser asks a DNS server for the IP address matching the domain name.
• The DNS server replies with the IP address of the web server.
• The browser reaches the web server through that IP address and requests the page.
• The web server sends the page data, and it is rendered on the user's screen.

• E-mail is a system for exchanging messages and data with specific users over the Internet using e-mail addresses.
• An address has two parts separated by @: the mailbox (user) name, then the mail server's domain, e.g. student@school.edu.eg.

Flow of a message

• The sender sends the message to his mail server with SMTP.
• That server passes the message on to the recipient's mail server.
• The recipient collects it with POP (download and delete from the server) or reads it with IMAP (view what is on the server, messages stay there and synchronise across devices).
• Attachments ride as encoded parts of the message, which is why size limits apply to the whole message.

Units and constants

• Transfer speed is the number of bits transmitted in one second, measured in bps.
• 1 Kbps = 1000 bps, 1 Mbps = 1000 Kbps, 1 Gbps = 1000 Mbps.
• 1 Byte = 8 bits; 1 KB = 1000 B, 1 MB = 1000 KB, 1 GB = 1000 MB.
• Transfer time (s) = data size in bits ÷ speed in bps; data size = speed × time; and actual speed = connection speed × transfer efficiency.

Worked example 1

• A file of 20 MB over a 40 Mbps link: size = 20 × 1000 × 1000 × 8 = 160,000,000 bits; speed = 40 × 1000 × 1000 = 40,000,000 bps; time = 160,000,000 ÷ 40,000,000 = 4 seconds.

Worked example 2

• A 1 GB file over a 100 Mbps link whose efficiency is 0.8: actual speed = 100 × 0.8 = 80 Mbps = 80,000,000 bps; size = 1 × 1000 × 1000 × 1000 × 8 = 8,000,000,000 bits; time = 8,000,000,000 ÷ 80,000,000 = 100 seconds.
• Always convert megabytes to megabits by multiplying by 8 — forgetting that factor is the classic error.

• Cloud computing is using the Internet to reach files, software and data processing without needing heavy local hardware.
• Advantages: access from anywhere and any device, no purchase or maintenance of servers, automatic backup and recovery, on-demand capacity and pay-per-use cost, and easy sharing and simultaneous editing.
• Points of care: the service needs connectivity, speed depends on the provider, and the data is held by someone else — so encrypt sensitive files and read the terms.
• Service models to recognise: infrastructure over the Internet, the platform on which programs are built, and finished software used through a browser.

• Generative AI performs mental and cognitive tasks that imitate what a human does; using it well improves workflow and raises productivity.
• How it works: the model is trained on enormous amounts of text and learns the statistical relations between words, then completes new requests by predicting what fits the context.
• It generates and edits text, translates, summarises, answers questions, produces images, and writes and debugs program code.
• In learning it acts as a tutor: it explains a difficult concept in simpler words, suggests research ideas, and turns a lesson into questions for self-testing.
• In programming it can produce a first draft, find the error in a snippet and explain the fix — the programmer still verifies and understands the result.

• A prompt is the instruction you give the model; the quality of the output depends on the clarity of the input.
• A good prompt fixes five things: the role, the context, the task, the constraints (length, level, language) and the output format (table, bullets, code block).
• Improve by iteration: give feedback on the first answer and ask for a revised version rather than repeating the same request.
• Verify the result: generative models can hallucinate — stating wrong facts and inventing references — so cross-check against trusted sources, and never submit generated work as your own without disclosure.