Topologies, Protocols and Web Services

A computer network is two or more computing devices connected together so they can exchange data and share resources such as files, printers and an internet connection. Each device on a network is called a node.

The standard way to classify networks is by geographical spread. This is not just a naming exercise. As the area grows, three things change together, and the board expects you to know why:

• The owner changes. A small network is owned by one person or one organisation. A large one has to cross public land, so it must rent lines from telecom companies.
• The speed usually falls, because signals travel further and pass through more intermediate devices.
• The cost and error rate rise, for the same reason.

PAN (Personal Area Network) is the smallest, organised around a single person, typically within about 10 metres. Your phone connected to wireless earbuds, a smartwatch syncing to that phone, or a laptop using the phone's hotspot — all PANs. The usual technologies are Bluetooth, infrared and USB. It is owned by the individual and needs no administrator.

LAN (Local Area Network) covers one building or one campus — a school computer lab, an office floor, a college hostel. Its defining feature is that the organisation owns the cabling itself, so it can install as much bandwidth as it likes. That is why a LAN is the fastest and most reliable of the four, and why file transfer inside your lab feels instant. Technologies are Ethernet (wired) and Wi-Fi (wireless).

MAN (Metropolitan Area Network) spans a city, of the order of tens of kilometres. It is normally built by joining several LANs together. Classic examples are a cable TV network across a city, and a bank connecting all its branches within Pune to one another. A MAN usually belongs to a single large organisation or to an Internet Service Provider, and it often has to use public rights of way, so it is costlier per node than a LAN.

WAN (Wide Area Network) spans countries and continents. The Internet is the largest WAN in existence. No single body owns it — it is thousands of networks agreeing to interconnect. It runs on leased telephone lines, undersea optical fibre and satellite links. Because data crosses many routers belonging to many operators, a WAN has the highest delay and the highest error rate of the four.

Worked example — classify each situation and justify it. The board asks this as a one-line scenario, and the marks are for the reason, not the label.

Worked example — watching the WAN actually work. Domain names were resolved to IP addresses from a machine in India. Each of these servers sits in a different network, and every one of them is reached over the WAN:

cbseacademic.nic.in    -> 164.100.230.217
www.nic.in             -> 23.58.95.96, 23.58.95.125
www.google.co.in       -> 142.251.221.99

Look at the middle line. One name answered with two addresses at once. That is not an error: a busy site runs several servers and publishes an address for each, so visitors are spread across them. If you run this lookup yourself, the numbers for the large sites will very likely differ from the ones printed here, because they change over time and from place to place. The name is stable; the address behind it need not be — which is exactly why we type names instead of numbers.

Network topology is the arrangement or layout in which the nodes of a network are physically connected to one another. Choosing a topology is really a decision about a trade-off: how much cable am I willing to buy, and how much of the network am I willing to lose when one thing breaks? Keep that sentence in mind — it answers almost every topology question.

Your syllabus lists only Bus, Star and Tree. Ring and Mesh topologies exist and you may have seen them in older books, but they are not part of the CBSE 2025-26 Unit 2 scope, so do not spend answer time on them.

Bus topology connects every node to a single common cable called the backbone or bus. A node sends its signal onto the backbone and the signal travels in both directions; every node sees it, but only the node with the matching address accepts it. Both ends of the backbone carry a terminator, a small resistor that absorbs the signal when it reaches the end. Without terminators the signal bounces back down the cable and collides with new data — this is why a missing terminator is a favourite exam detail.

• Advantages: uses the least cable, so it is the cheapest to install; simple to understand; easy to extend a small network by joining another length of cable.
• Disadvantages: a break anywhere in the backbone brings down the whole segment; faults are hard to locate because there is no central point to test from; only one node can transmit at a time, so performance drops sharply as nodes are added.

Star topology connects every node by its own dedicated cable to a central device — a hub or, in any modern network, a switch. Nothing travels directly between two nodes; everything passes through the centre. This is by far the most common topology in real school and office LANs.

• Advantages: if one node's cable breaks, only that node goes offline and the rest keep working; adding or removing a node does not disturb anyone else; fault finding is easy because every cable ends at one place; the switch can send data only to the intended node, so several conversations run at once.
• Disadvantages: needs much more cable than a bus; the central device costs money; and the central device is a single point of failure — if the switch dies, the entire network dies.

Tree topology (also called hierarchical topology) is a combination of bus and star. Several star networks are connected to a common backbone cable, forming levels: a root node at the top, then branches, then the nodes. A school with one switch per lab, all the lab switches wired back to a main school switch, is a tree.

• Advantages: scales well, because a whole new branch can be added without touching existing branches; suits organisations that are already divided into groups such as blocks or departments; a fault inside one branch is isolated to that branch.
• Disadvantages: needs the most cable and is the most complex to configure; and it depends on the backbone — if the backbone between branches fails, the branches keep working internally but can no longer talk to each other, splitting the network into islands.

Worked example — putting a number on the trade-off. Take one lab of 20 computers standing in a row, 3 m apart, with a 2 m drop from each desk down to the cabling, and the switch standing at one end of the row. Compare the cable a bus needs against the cable a star needs, and compare what one broken cable costs you:

nodes = 20, spacing = 3 m, drop = 2 m, switch at one end of the row

BUS  : backbone 19 x 3 = 57 m  +  20 drops x 2 m  =  97 m
STAR : 20 separate runs back to the switch        = 610 m
STAR uses 6.3x the cable of BUS

one backbone break in BUS    -> up to 20 nodes affected
one drop-cable break in STAR -> 1 node affected
central switch fails in STAR -> 20 nodes affected

Read the two halves together, because that is the whole lesson. The star costs about six times the cable, and in exchange a broken cable takes down one machine instead of twenty. Schools and offices happily pay that six times, because cable is bought once but a dead lab costs a period every time. The one risk the star does not remove is the switch itself, which is why the switch is the thing you keep a spare of.

One practical note on that 610 m: it assumes the switch sits at one end of the row, which is the worst case. Move the switch to the middle of the row and the same twenty machines need only about 340 m. Where you put the central device changes the bill substantially, which is why wiring cupboards are placed centrally in real buildings.

A protocol is a set of rules that governs how data is exchanged between devices on a network. Both sides must follow the same rules — the format of the message, the order in which messages are sent, and what to do when something goes wrong. Two computers with perfectly good cabling between them cannot communicate if they do not agree on a protocol.

A port number is a 16-bit number that identifies which program on a machine a message is meant for. The IP address gets the data to the right computer; the port number gets it to the right application on that computer. Well-known services use fixed port numbers so a client knows where to knock without being told.

Your syllabus names nine protocols. Learn each one's full form, its job, and its port where it has a standard one.

1. HTTP — HyperText Transfer Protocol (port 80). The rules by which a web browser asks a web server for a page and the server replies. It works on a request-response cycle and is stateless: the server does not remember you between two requests, which is why websites need cookies or sessions to keep you logged in.

2. HTTPS — HyperText Transfer Protocol Secure (port 443). This is ordinary HTTP wrapped inside an encryption layer called SSL/TLS. The content is the same; what changes is that anyone intercepting the traffic sees only scrambled bytes. The server also presents a digital certificate issued by a Certifying Authority, which proves it really is the site it claims to be. The padlock in the address bar means both of these succeeded. Use HTTPS for anything involving passwords, banking or personal data.

3. FTP — File Transfer Protocol (port 21 for control, port 20 for data). Used to upload and download files between a client and a server — for example, a web developer copying finished pages up to a hosting server. FTP is unusual in using two connections: commands such as 'list this folder' travel on port 21, while the file contents travel separately on port 20. Anonymous FTP lets the public download files without an account.

4. SMTP — Simple Mail Transfer Protocol (port 25). The protocol for sending mail: it pushes a message from your mail client to your mail server, and onward from one mail server to another. Remember it as a push protocol — it moves mail away from you.

5. POP3 — Post Office Protocol version 3 (port 110; port 995 when encrypted). The protocol for receiving mail: it pulls messages from your mail server down to your device. In its default behaviour POP3 downloads the mail and deletes it from the server, so the message then exists only on that one device. Remember it as a pull protocol.

6. TELNET (port 23). Lets you log in to a remote computer over the network and use it through a text terminal, as though you were sitting at it. Its fatal flaw is that it sends everything — including your password — as plain, unencrypted text, so anyone capturing the traffic can read it. For this reason TELNET has been replaced in real use by SSH, which does the same job with encryption.

7. PPP — Point-to-Point Protocol. A data link layer protocol that connects exactly two nodes over a direct link, such as a dial-up modem line or a DSL connection. It handles authentication (checking your username and password with the ISP), error detection, and carrying several different network-layer protocols over the one link. Because PPP works below the transport layer, it has no port number at all — port numbers only exist for TCP and UDP. This is a favourite trick question.

8. TCP/IP — Transmission Control Protocol / Internet Protocol. Not one protocol but the suite the entire Internet is built on, so it has no single port of its own. The two halves do different jobs:

• TCP is connection-oriented and reliable. It breaks a message into numbered packets, sends them, waits for an acknowledgement for each, retransmits anything lost, and reassembles the packets in the correct order at the other end.
• IP handles addressing and routing — giving every machine an IP address and deciding which path each packet takes. IP is connectionless and best-effort: it will try to deliver a packet but makes no promise, which is exactly why TCP's checking is needed on top.

9. VoIP — Voice over Internet Protocol. A technology that carries voice as digital packets over an IP network instead of over the traditional telephone circuit. Your voice is sampled, digitised, compressed and sent as packets. WhatsApp calls, Google Meet and Skype all use VoIP. It is much cheaper than a conventional trunk call because it uses the internet connection you already pay for. VoIP has no single fixed port: call setup commonly uses SIP on port 5060 (5061 when encrypted), while the audio itself flows over RTP on ports chosen dynamically for each call.

Worked example — the ports are real, so let us knock on them. Connecting to each service on its standard port makes the server answer in its own protocol. These are the actual replies received:

FTP control     ftp.gnu.org:21    -> 220 GNU FTP server ready.
SMTP            smtp.gmail.com:25 -> 220 smtp.gmail.com ESMTP ... gsmtp
POP3 over SSL   pop.gmail.com:995 -> +OK Gpop ready for requests
POP3 plain      pop.gmail.com:110 -> blocked / no reply

Notice three things. FTP and SMTP both greet with the code 220, meaning 'service ready' — that numeric-reply style is part of what makes them protocols rather than free conversation. POP3 answers +OK instead, its own way of saying the same thing. And plain POP3 on port 110 gave no reply at all while the encrypted version on 995 worked: major mail providers now switch the unencrypted port off, the same shift that pushed HTTP towards HTTPS and TELNET towards SSH.

Worked example — HTTP is readable, HTTPS is not. An HTTP request was sent to a real web server on port 80. Both the request and the reply cross the network as ordinary text that anybody in between could read:

GET / HTTP/1.0
Host: example.com

HTTP/1.1 200 OK
Date: Fri, 21 Aug 2026 15:21:41 GMT
Content-Type: text/html
Connection: close
Server: cloudflare

The same kind of request was then made over HTTPS on port 443. Before any web data moved, the two sides negotiated encryption and the server proved its identity with a certificate:

TLS version negotiated : TLSv1.3
cipher                 : TLS_AES_128_GCM_SHA256
certificate issued to  : www.python.org
certificate issued by  : GlobalSign nv-sa

HTTP/1.1 200 OK

The final line is identical in both cases — HTTP/1.1 200 OK — which proves the point that HTTPS is not a different language. It is the very same HTTP, carried inside a TLS envelope, with a certificate from a Certifying Authority (here GlobalSign) vouching for who the server is.

The World Wide Web (WWW) is a vast collection of interlinked documents and resources, identified by URLs and accessed over the Internet using HTTP. It was proposed in 1989 by Tim Berners-Lee at CERN, the European particle physics laboratory.

The single most commonly examined idea here is this: the Internet and the WWW are not the same thing. The Internet is the physical network — the cables, fibre, routers and satellites, plus the addressing that ties them together. The WWW is one service that runs on top of that network. Email, VoIP calls and file transfer are other services on the same Internet, and none of them is the Web. So: the Internet is the road; the WWW is one kind of traffic on it.

What makes the Web a web is hypertext — text containing links that jump you to another document anywhere in the world. The documents are written in HTML.

HTML (HyperText Markup Language) is the language used to create web pages. It is a markup language, not a programming language: it has no variables, no loops and no decisions. It works by wrapping content in tags that tell the browser what each piece of content is and how to present it. Its defining features:

• The tag set is fixed and predefined. You may use only the tags the standard defines — p, table, h1 and so on. You cannot invent student, because no browser would know what to do with it.
• It is not case-sensitive.
• It is forgiving. Leave a tag unclosed and the browser guesses your intent and displays the page anyway.
• Its purpose is presentation — how information should look to a human reader.

XML (eXtensible Markup Language) looks similar but exists for the opposite purpose. XML is used to store and transport data, especially between two different systems that must agree on the data but may be written in different languages and run on different platforms. Its defining features:

• Tags are user-defined. This is what 'extensible' means — you invent tags that describe your data, such as student, marks or city. The document becomes self-describing.
• It is case-sensitive: Name and name are two different elements.
• It is strict. An XML document must be well-formed or it is rejected outright: exactly one root element, every opening tag closed, tags properly nested, and every attribute value in quotes.
• Its purpose is data, not display. XML by itself says nothing about fonts or colours.

Worked example — the same two files, side by side. First an HTML page that displays a mark list. Every tag in it is a standard HTML tag:

<!DOCTYPE html>

Class XII-A Marks

  Class XII-A - Computer Science
  <table>
    <tr><th>Roll</th><th>Name</th><th>Marks</th></tr>
    <tr><td>12</td><td>Ananya Iyer</td><td>92</td></tr>
    <tr><td>17</td><td>Rohit Deshmukh</td><td>85</td></tr>
  </table>
  <p>Total: 177</p>

Listing the tags this file actually uses confirms they are all borrowed from HTML's fixed vocabulary — none was invented:

tags used in marks.html: ['b', 'body', 'h2', 'head', 'html', 'p',
                          'table', 'td', 'th', 'title', 'tr']

Now the same information as XML. Notice that class, student, name, city and marks are tags invented for this data — they exist nowhere in any standard:

<?xml version="1.0" encoding="UTF-8"?>

  
    Ananya Iyer
    Chennai
    92
  
  
    Rohit Deshmukh
    Pune
    85
  

Because those tags describe the data, a program can pull meaning straight out of the file. Reading the XML above and extracting the fields gives:

root tag  : class | attribute name = XII-A
roll 12: Ananya Iyer, Chennai, CS = 92
roll 17: Rohit Deshmukh, Pune, CS = 85
total marks: 177

The HTML file contains the same marks, but they are trapped inside td cells that mean only 'a box in a table'. The XML file says which number is a mark and whose it is, so the total could be computed automatically. That is the real difference: HTML describes appearance, XML describes meaning.

Worked example — proving XML is strict and HTML is not. Take one badly written fragment, with p never closed and a br that has no closing tag, and feed it to both parsers. As XML it is fatally rejected:

input: <p>Ananya Iyer<br>Chennai

XML  : ParseError: no element found: line 1, column 25
HTML : start tag: p
       text     : Ananya Iyer
       start tag: br
       text     : Chennai

Same characters, opposite outcomes. The browser shrugs and shows the text; the XML parser refuses to produce anything at all. Case sensitivity behaves the same way — closing Name with /name is fine in HTML but fatal in XML:

HTML:     -> all three read as the same tag: body
XML : Ananya
      -> ParseError: mismatched tag: line 1, column 23

This strictness is a feature, not a nuisance. XML is meant to be exchanged between machines, and a machine that guesses at malformed data will silently produce wrong answers. Better to fail loudly.

Every computer on the Internet is identified by an IP address, a number such as 164.100.230.217. Numbers are fine for machines and hopeless for people, so we use domain names instead — readable names like cbseacademic.nic.in.

The Domain Name System (DNS) is the service that translates a domain name into its IP address. It runs on port 53. When you type a name into a browser, the browser first asks DNS for the matching IP address, and only then can it contact the server. This lookup is invisible but it happens before every single page you open.

Worked example — a real lookup. Resolving a name returns the address the browser must actually connect to:

lookup : cbseacademic.nic.in
answer : 164.100.230.217

the browser now opens its connection to 164.100.230.217, never to the name itself

Domain names are hierarchical, and you read them right to left, from most general to most specific. Take cbseacademic.nic.in:

• in is the Top Level Domain (TLD), the rightmost part.
• nic is the second level domain, the registered name.
• cbseacademic is a sub-domain of that.

TLDs come in two kinds. Generic TLDs (gTLD) describe the type of organisation: .com commercial, .org organisation, .net network, .edu education, .gov government, .info information. Country code TLDs (ccTLD) identify a country in two letters: .in India, .uk United Kingdom, .jp Japan, .us United States. India's .in domain is administered by the National Internet Exchange of India (NIXI), and you will often meet its combined forms — .co.in for companies, .ac.in for academic institutions, .gov.in and .nic.in for government bodies.

A domain name gets you to a server. To point at one exact resource on that server you need a URL (Uniform Resource Locator) — the complete address of a specific file or page on the Web.

Worked example — breaking a URL into its parts. Take this URL and split it. These are the components an actual URL parser reported for it:

https://www.vidyalaya.edu.in:8080/class12/cs/notes.html?unit=2&topic=networks#protocols

scheme   : https
hostname : www.vidyalaya.edu.in
port     : 8080
path     : /class12/cs/notes.html
query    : unit=2&topic=networks
fragment : protocols

Each part has a name and a job, and the board asks you to identify them:

A website is a collection of related web pages kept under a single domain name, linked to one another and usually reached through a home page. A website is static if every visitor is served the same stored pages, and dynamic if pages are built when requested, so different visitors see different content — your own marks on a results portal, for instance.

A web browser is the client software on your device that requests pages and renders them. It sends the HTTP request, receives the HTML, CSS and JavaScript, interprets them, and draws the page. Common browsers are Google Chrome, Mozilla Firefox, Microsoft Edge, Apple Safari and Opera.

A web server is the server side of that exchange: software (running on a powerful, permanently connected computer) that stores the website's files and delivers them over HTTP or HTTPS whenever a browser asks. Common web server software includes Apache HTTP Server, Nginx and Microsoft IIS. Do not confuse the two — the browser requests, the server responds.

Worked example — web servers name themselves. Every HTTP response can carry a Server header saying which software produced it. Three different real sites were contacted, and they replied:

example.com      -> Server: cloudflare
www.python.org   -> server: nginx
ftp.gnu.org      -> Server: Apache/2.4.52

That is direct evidence of the client-server model: three separate machines running three different web server programs, all answering the same HTTP request that any browser would send.

Web hosting is the service of renting space and bandwidth on a web server so that your website stays available on the Internet 24 hours a day. You could in principle run a server from home, but it would need a permanent connection, a fixed IP address, backup power and constant security patching — so almost everybody rents instead. The main types, cheapest first:

• Shared hosting — many websites share one physical server and its resources. Cheapest, and fine for a school or small business site, but a busy neighbour can slow you down.
• VPS (Virtual Private Server) hosting — one physical server is divided into isolated virtual machines, each with guaranteed resources. Costlier, but a neighbour's traffic cannot affect you.
• Dedicated hosting — an entire physical server is rented by you alone. Maximum control and performance, highest cost.
• Cloud hosting — the site runs across a pool of servers and can grow or shrink with demand, so a sudden rush (a results day, say) is absorbed automatically.
• Colocation — you own the server hardware but keep it in a provider's data centre for their power, cooling and connectivity.