Networks and the Internet

A phone sending a photo and a classroom sharing a printer both use a network. A computer network is two or more connected devices that exchange data. The devices do not have to be computers: printers, sensors, phones and game consoles can all take part.

Devices, links and shared resources

Each connected device is a node. A connection between nodes is a link, which may use cable or radio signals. A device can use several links during one task. A phone might send data by radio before cables carry it farther.

Networks make sharing possible. A school can keep files on a shared server, let many classrooms send work to one printer, and give hundreds of devices access to one internet connection. Students can also send messages and work together from different rooms.

Shared equipment brings shared problems. Someone must buy, set up and maintain it. If a school's internet link fails, many classes lose online access together. Connecting devices also creates routes that unwanted users might try to use, so access needs protection.

Agreed rules make messages work

A protocol is a set of agreed rules for communication. It can describe the shape of a message, the order of replies and what happens after an error. Hardware and software follow protocols; the protocol itself is the rule set.

On a two-way radio, saying "over" tells the other speaker it is their turn. Saying "say again" asks for a repeat. Both speakers know the meanings. Without a shared rule for taking turns, they might speak together and hear neither message clearly.

Computer protocols divide up the work:

One visit to a website can use all these rules. A device does not choose just one protocol for everything it does.

A network can work without the internet

Imagine two computers and a printer joined through a switch. The computers can send print jobs across those links. This is already a network, even if it has no internet connection.

Now add a shared file server and an internet connection. If the internet connection breaks, the local cables and devices still work. Printing and opening files from that local server can continue. A file stored only on a faraway cloud server needs the internet link, so it becomes unavailable.

Passing messages along is an old idea

In 1794, a line of 15 semaphore stations connected Paris and Lille in France. Operators watched the previous tower's arm positions through telescopes, then copied them for the next tower. The signal code supplied shared rules. Each station passed a message onward, an idea that also appears in today's networks when routers pass data toward its destination.

A watch talking to a phone and a bank linking distant branches are both networks. Their different reach gives us useful names: PAN, LAN and WAN. Look at the area covered and who runs the connections when choosing a name.

Three useful scales

A PAN can use Bluetooth or a short cable. For example, plugging a phone into a laptop by USB can create a personal connection. The name describes the network's scale, not one particular wireless technology.

A LAN may join computers in several buildings on one school site. The school usually manages its switches, access points and cables. Inside the same LAN, one device can reach shared resources without sending every message across the internet.

A WAN joins networks over larger distances, such as between cities or countries. Telecommunications companies usually provide the long links. The internet joins networks worldwide and is often described as the largest WAN.

Follow one song through three networks

A music service sends a song from a distant server to a home. Across the internet, it travels through a wide network of linked networks. Inside the home, Wi-Fi carries it to a phone on the LAN. Bluetooth then carries the sound from the phone to earbuds on a PAN.

The same task crosses all three scales. Calling the earbud connection a PAN does not make the home network stop being a LAN.

Two clues that can mislead

Device count is not the deciding clue. A home LAN may contain five devices, while a university LAN contains thousands. A WAN linking a few distant offices can have fewer devices than that university LAN.

Wireless does not mean WAN. Wi-Fi in one house is a wireless LAN, sometimes called a WLAN. "Local" describes where the network operates. "Wireless" describes how a link carries data. To classify a diagram, first find the connected area; then describe its wired or wireless links separately.

Different pieces of hardware do different jobs along a message's route. Some connect a device, some join local links, and others move data to another network. A single home box can combine several of these jobs.

Join a device to the LAN

A network interface card (NIC), or network adapter, gives a device its network connection. A wired adapter has a cable socket. A wireless adapter sends and receives radio signals using an antenna. Phones and laptops usually have adapters built in.

A switch joins wired devices inside a LAN. It learns which device can be reached through each port, then forwards data through the useful port. An older hub copied incoming data to every port, wasting capacity on devices that did not need it.

A wireless access point joins wireless devices to the LAN. It exchanges radio signals with laptops and phones, and normally connects by cable to the rest of the network. Several access points give a large school coverage across its rooms.

Reach other networks

A router passes packets between networks. It reads each packet's destination IP address and forwards it toward that destination. A school's internet router connects its LAN to the provider's network.

A modem converts between the signals on the provider's line and the local network's signals. DSL telephone lines and cable TV lines use modems. Fiber connections use a similar connection box called an optical network terminal (ONT).

Protect or share a connection

A firewall checks traffic against rules and blocks traffic the rules reject. Firewalls can be software, hardware or both. Most computers and home routers include one.

A mobile hotspot shares a cellular connection through Wi-Fi. A laptop joins the phone's Wi-Fi network, and the phone sends its internet traffic through the cellular network. "Hotspot" can also mean a public place offering Wi-Fi.

Trace two classroom tasks

For a wireless laptop printing to a wired printer on the same LAN, follow the path:

1. The laptop's adapter sends radio signals to the access point.
2. A cable carries the data from the access point to a switch.
3. The switch forwards it through the port leading to the printer.

The data stays local, so crossing to another network is unnecessary. For a website, the switch also passes the outgoing data toward the router, which forwards it to the internet provider.

The home device called a "router" often contains a router, a small switch, an access point and a firewall. It may also contain a modem. That explains its many sockets and settings: one case holds several separate functions. The modem's line conversion and the router's forwarding remain different jobs.

Data travels as bits, represented by signals on a link. Copper cable uses electrical signals. Fiber-optic cable uses pulses of light in thin strands of glass or plastic. The choice affects distance, capacity and installation.

Copper for short links

Common Ethernet cable contains twisted pairs of copper wires. Twisting helps the pairs work together in a practical network cable. A typical four-pair copper Ethernet link reaches about 100 meters. Copper is affordable, flexible and easy to connect, making it useful between desks and switches.

The signal weakens over distance. Electrical interference from motors or power lines can also disturb it. A damaged cable or loose connector can interrupt an otherwise suitable short link.

Coaxial cable is another copper cable. It has a central wire surrounded by shielding. Many providers use it to bring cable TV and internet service into homes.

Fiber for long, busy links

Fiber turns bits into light pulses and back again at the receiving end. The strand guides the light along its length. Fiber carries large amounts of data over many kilometers before the signal needs boosting. Electrical interference does not disturb the light signal.

Installing fiber costs more, and joining its ends requires care. Sharp bends can damage it. These costs are worthwhile for routes carrying traffic between whole buildings, cities or countries.

Match the cable to the route

A school needs a link between buildings 800 meters apart. A single usual copper Ethernet run cannot cover that distance. Fiber fits the distance and can carry the combined traffic of many classrooms.

A desktop computer sits 8 meters from its switch. Copper fits comfortably within its distance limit and is simpler to install. Choosing fiber for every desk would add installation cost without a need for its long reach.

Undersea fiber carries almost all intercontinental data. A message sent from a wireless phone therefore often spends most of its journey inside cables.

Capacity and travel time are different

Signals in copper and fiber travel at roughly 200,000 kilometers per second. Fiber's main advantages are its capacity and reach. It does not remove the delay caused by distance.

For a 5,600 km undersea route, the signal's travel time is:

$5600÷200000=0.028$ seconds, or 28 ms.

The outward and return journeys need at least 56 ms in total. Equipment and waiting add more delay. Increasing how much data the cable carries each second does not shorten that 5,600 km journey.

Wi-Fi, Bluetooth and cellular connections all carry data using radio waves. They serve different distances and tasks. Choosing well means looking at movement, range, data needs and battery use.

Wi-Fi around a building

Wi-Fi links a device to an access point in a LAN. A network name identifies the Wi-Fi network, and a password usually controls joining. Indoors, coverage commonly reaches tens of meters. Walls, metal objects and distance weaken the signal.

Wi-Fi uses bands around 2.4 GHz and 5 GHz, with 6 GHz available on newer equipment. The 2.4 GHz band generally reaches farther through walls but is more crowded. Higher bands can provide greater speed over short distances. Other nearby networks and some devices, including microwave ovens near 2.4 GHz, can cause interference.

Devices on the same radio channel take turns using it. More devices can therefore mean more waiting even when the signal bars look strong.

Bluetooth around a person

Bluetooth suits short links, typically around 10 meters, with low power use. A keyboard, watch or earbuds can exchange modest amounts of data with a nearby device without using a building's Wi-Fi network. Devices are paired so they can recognize and reconnect to each other.

The low power use helps small batteries last. A watch sending step counts does not need a long-distance connection to a cell tower for that nearby transfer.

Cellular connections across an area

A cellular network divides its coverage into cells served by base stations, often called cell towers. A phone exchanges radio signals with a nearby station. As the phone moves, the network passes its connection between stations.

The stations connect to the phone company's wider network, mostly through cables. The data can then travel to the internet. Radio covers the phone's local part of the journey.

4G and 5G are generations of cellular technology. 5G offers more bandwidth, lower latency and capacity for more devices. Its fastest high-frequency bands have shorter reach and need more small cells. Available coverage and equipment affect what a person actually gets.

Read the signal bars carefully

Full Wi-Fi bars describe the link from the device to the access point. The access point can still work when the building's internet connection has failed. Nearby devices may keep sharing local files even though online services stop.

Weak Wi-Fi in one room suggests checking distance and obstacles. Strong Wi-Fi throughout the building with every internet service failing suggests looking farther along the connection. Signal strength is one clue about one link, not a report on the whole journey.

A connection should fit how a device is used. A server that stays in one place has different needs from a tablet carried between rooms. Compare speed, reliability, mobility, cost and security before choosing.

Compare the trade-offs

Wired links usually provide steady speed and low delay. They avoid nearby radio interference. A suitable cable does not lose signal because a classroom door closes or a user walks farther from the access point.

Wireless links make adding and moving devices easier. Running cables to every seat may be costly or impractical, especially in an old building. Thick walls still matter: avoiding a cable route does not make radio signals pass easily through stone.

Make the choice in steps

First list the task. Will the device download large files, join video calls or send a few sensor readings? Then ask whether it moves and how far it is from the connection point. Check installation cost and the protection needed for its data.

A desktop for online games sits beside the router. It stays still, and low delay matters, so a cable is a strong choice. A class carries tablets between rooms, so Wi-Fi supports their movement. A file server serves many users throughout the day, so a wired connection supports its need for reliable performance.

Mix links in one network

A school can connect its server, desktops and access points by cable. The access points then connect tablets and laptops by radio. Wireless use still depends on the wired links behind the access points.

Avoid treating "wired" as a guarantee of higher speed. A recent Wi-Fi connection close to its access point can outperform an old, slow wired link. Check the actual equipment and task. Similarly, a cable makes listening harder for someone outside the building, but accounts, permissions and other protections still matter once a person can access the network.

A connection can carry a large file quickly yet still feel slow during a game. Bandwidth, latency and reliability measure different things.

Bandwidth measures capacity

Bandwidth is the most data a link can carry each second. Mbps means millions of bits per second; Gbps means billions. Actual transfer speed is usually lower because other users share capacity and control information uses some of it.

File sizes often use bytes, while connection speeds use bits. One byte contains 8 bits. The capital B in MB means bytes; the lowercase b in Mb means bits.

To find a best-case transfer time:

1. Convert the file size from megabytes to megabits by multiplying by 8.
2. Divide by the speed in Mbps.
3. Give the answer in seconds.

For a 100 MB file at 40 Mbps:

$100\times 8=800$ megabits.

$800÷40=20$ seconds.

For a 2,000 MB update at 100 Mbps, the file contains 16,000 megabits. Dividing by 100 gives 160 seconds, or 2 minutes 40 seconds. Dividing megabytes straight by Mbps misses the conversion and gives a time eight times too small.

Latency measures delay

Latency is travel delay, usually measured in milliseconds. A round-trip measurement includes a small message going out and a reply coming back. Longer distances, more equipment and waiting in queues add delay.

Picture bandwidth as the number of lanes on a highway and latency as one car's journey time. Adding lanes helps more cars travel together, but leaves the road's length unchanged.

A satellite 36,000 km above Earth relays a signal up and back down. That one-way journey between ground points covers 72,000 km. At about 300,000 km per second, radio takes:

$72000÷300000=0.24$ seconds, or 240 ms.

A connection using that route can have ample bandwidth while a game's replies still arrive late.

Reliability measures consistency

A reliable connection delivers data consistently, with few errors, lost packets or drop-outs. A high advertised speed is less useful if the connection repeatedly stops.

Explain overload

Thirty streams at 5 Mbps each need 150 Mbps. A 100 Mbps link cannot supply that total. If shared equally, it gives about 3.3 Mbps per stream, so playback pauses while more data arrives. Queues grow and packets may be dropped.

Weak radio signals, old equipment and a busy remote server can also cause slowness. Compare several services and devices before deciding which part needs improvement.

A large photo does not cross a network as one solid object. The sending device splits its data into packets, small blocks that travel as separate units. The receiving device uses their ordering information to rebuild the data.

What travels in a packet?

The payload carries part of the actual message, picture or sound. The header carries control information, including addresses that identify the sender and destination. For transfers that need ordered delivery, numbering information identifies where data belongs. Check information helps detect damaged data.

All these parts are bits. A header containing an address and a payload containing music both travel as electrical signals, light pulses or radio signals, depending on the link.

Think of mailing a long letter on numbered postcards. Each card carries an address and one part of the letter. The reader uses the numbers to restore the order. The number tells where the piece belongs; it is not another copy of the whole letter.

Count the packets

A packet's stated data capacity is the space available for payload. For a 4,000-byte file with at most 1,500 data bytes per packet:

• The first packet holds 1,500 bytes.
• The second holds another 1,500 bytes.
• The remaining 1,000 bytes need a third packet.

Dividing gives more than two packets, so round up to three. Rounding down would leave data unsent. A 3,000,000-byte photo with the same payload capacity divides exactly into 2,000 packets.

Put the pieces back in order

Split the message "MEET ME AT THE LIBRARY AT 4" into blocks of at most 10 characters, counting spaces:

These are 10, 10 and 7 characters, for 27 altogether. The second payload starts with a space. If the packets arrive as 2, 3, 1, the receiver waits and places them in the correct order before delivering the complete text. Routers along the way forward packets; they do not rebuild the whole message at each stop.

Share links fairly and recover smaller pieces

Packet switching lets packets from different users take turns on the same link. A short message can travel between parts of a large video transfer. A traditional landline call reserved a connection throughout a call, including during silences; packet switching shares capacity as data arrives.

If one piece is lost during a reliable transfer, that piece can be sent again. The sender does not need to restart the whole file. If a link fails and another path exists, later packets can use it.

Packets from one file can follow different paths and arrive out of order. They can also follow the same path while conditions stay steady. The destination must use the ordering information, not rely on whichever piece arrives first.

A packet may cross many networks before arriving. At each junction, a router reads its destination IP address and chooses the next step. The routers work together without one central machine directing every journey.

Each router chooses a next hop

A routing table tells a router which next link leads toward different destination networks. A hop is one step between routers. Each router chooses the next hop; it does not need to plan every later step.

Road signs give a useful picture. At each junction, a driver chooses the next road toward a destination. The route follows the roads available, not a straight line across a map. Internet routes also depend on available links and network rules.

Routers exchange information about networks they can reach. After a link fails, routing information can change so later packets travel another way. This takes an alternative path: a building with just one outside cable cannot route around a cut in that cable.

Work through a route change

Consider links with these delays:

For this network, suppose the chosen route has the lowest total link delay. Add the links on each possible route without repeating a router:

• A–B–D takes 15 ms.
• A–C–B–D takes 17 ms.
• A–C–D takes 24 ms.
• A–B–C–D takes 28 ms.

The first route wins. If B–D fails, both routes using that link disappear. A–C–D is then the best remaining route at 24 ms. Data still arrives, but the trip takes longer.

Real routing follows a network's rules and costs. A busy link usually remains in use: packets wait in queues or get dropped. Congestion does not by itself mean routers select a quieter route.

Redundancy makes resilience possible

Redundancy means having extra paths or copies. Resilience is the ability to keep working after a failure. Multiple connected paths give the internet resilience when one link or router fails.

Extra paths cannot replace a missing destination. If the only server holding a website fails, several good routes to it do not restore the site. Keeping another working copy addresses that different weakness.

On January 15, 2022, an eruption and tsunami broke Tonga's international undersea cable. Limited satellite links provided some contact until the cable was repaired on February 22, about five weeks later. A local network can keep working during an outside outage, but card payments, messaging and cloud-only files may become unavailable.

Stop packets circling forever

A packet carries a hop limit. Each router lowers it by one; when the limit is used up, the packet is discarded. This prevents a routing loop from sending the same packet around indefinitely. Packet ordering at the destination handles pieces that take different routes; the hop limit handles pieces that fail to make progress.

Data can arrive changed, or fail to arrive at all. A transmission error changes received data. Packet loss leaves a missing piece. Understanding the cause helps you choose a useful fix.

Signals can change bits

Electrical or radio interference, weak signals, damaged cables and faulty hardware can disrupt signals. Even one flipped bit matters. In ASCII text, the letter A is 01000001. Flipping the second-to-last bit produces 01000011, the code for C.

A check value helps the receiver detect errors. The sender calculates a value from the data and sends it along. The receiver calculates again and compares the results. A mismatch shows that the data changed.

For a simple addition check, data values 12, 7 and 30 give 49. If 12, 9 and 30 arrive, their sum is 51. That differs from the sent value, so the receiver rejects the damaged data. A matching simple sum can still miss some changes, such as two values swapping places.

Queues can overflow

A router stores waiting packets in a queue. If packets enter faster than the next link can send them, the queue grows. When memory is full, the router drops packets it cannot store. This is congestion. Bad wireless signals and broken equipment can also cause loss.

Reliable transfers track what arrived

IP makes a best effort to deliver packets, but it does not guarantee arrival. TCP adds reliable, ordered delivery. It tracks numbered data and uses acknowledgments, replies reporting received data. Missing data can then be sent again.

Imagine a transfer divided into pieces 1 to 5, with piece 3 lost. The receiver keeps the pieces that arrived. Delivery information lets the sender detect the gap and resend the missing data. Once the gap is filled, the receiver delivers the complete data in order.

The numbers are ordering information for the data. TCP can acknowledge a continuous stretch of received data, so a separate reply for each numbered packet is unnecessary.

Resending affects what people see

A file download needs every piece. Waiting for a missing piece makes it slower, even if the finished file looks correct. Sending the entire file again would waste the pieces already received.

A live call has a different need. An old sound fragment may be useless by the time it arrives. Live calls and games often favor timely data over resending every lost piece. Loss can then appear as a sound gap, frozen picture or sudden jump. Adding bandwidth helps congestion, but a damaged cable still needs a hardware fix.

An IP address identifies a device's connection on a network. Packet addresses let data reach its destination and let replies find their way back. The address belongs to a network connection and can change when the device moves.

Recognize IPv4

An IPv4 address has four numbers separated by dots, such as 192.0.2.25. Each number uses 8 bits and ranges from 0 to 255. Four groups of 8 bits make a 32-bit address.

To check its written form, count four parts and check each part's range. 198.51.100.7 has the correct form. 192.168.1.300 fails because 300 is too large. 10.0.5 has only three parts.

IPv4 has about 4.3 billion possible values:

Some values are reserved for special uses. The central pool of unused IPv4 addresses ran out in February 2011, helping drive the need for a larger address system.

Recognize IPv6

IPv6 uses 128-bit addresses. They are written as groups of hexadecimal digits, using 0–9 and a–f, separated by colons. In 2001:db8::1, the double colon shortens a run of zeros.

IPv6 has about $3.4\times {10}^{38}$ possible values. IPv4 and IPv6 operate alongside each other. A longer address gives more possible connections their own addresses; it does not by itself make a particular download faster.

Separate private and public addresses

Private addresses work inside local networks and are not directly reachable across the public internet. The private IPv4 ranges are:

• Addresses beginning 10.
• Addresses from 172.16. through 172.31.
• Addresses beginning 192.168.

Different homes can reuse the same private address. Inside one network, two active devices using the same address cause a conflict.

A home router commonly shares one public address among local devices. Suppose a laptop is 192.168.1.20 and a phone is 192.168.1.21. To outside servers, their requests can appear to come from the router's shared public address. The router tracks which internal device should receive each reply.

The address 203.0.113.50 can stand in for that public address in a diagram. It is reserved for documentation; real public connections use other addresses.

Addresses can change

Devices usually receive an address automatically when they join a network. The router often provides this service. A phone can have different addresses at school, at home and on cellular service. A printer or server often keeps a fixed address so other devices can find it consistently.

An IP address identifies a connection, not the user's name. It can indicate the network and a rough area. The provider may link it to a customer account, but reading an address alone does not tell you who is holding a device.

People type readable website names, while packets use number addresses. A URL identifies a web resource. DNS, the Domain Name System, helps turn its host name into an IP address so the request can travel.

Read the address in parts

Consider https://www.example.org/news/today.html:

The complete string is the URL. The domain example.org is the readable registered name within the host name. Read its parts from right to left: org is the top-level domain, example is registered within it, and www identifies a host in that domain. Other endings include com, edu, nl and au.

Changing the path can request another page from the same host. Changing the host can require a different address lookup. DNS finds a host's address, not the file named by the path.

Follow a lookup

DNS is a distributed directory, spread across many servers. A device first checks for a saved answer. If none is available, it asks a resolver, often supplied by the internet provider.

If that resolver also lacks a saved answer, it follows a chain:

1. A root server directs it toward servers for the top-level domain.
2. Those servers direct it toward the servers responsible for the domain.
3. The responsible server returns the requested host's IP address.
4. The resolver passes that answer back to the device.

For www.example.org, the chain moves through the root, org, and example.org. In a worked diagram, 203.0.113.10 can represent the returned address. This reserved documentation address shows the role of an answer without giving a live destination.

The browser then uses the IP address to contact the web server. The web server answers with the requested page. A root server usually supplies a direction to the next server, not the final answer for every host.

Save answers for later

Caching means keeping an answer for a set time. A device or resolver with a recent answer can avoid repeating the full chain. This speeds repeated visits and reduces work for DNS servers.

Caching also explains a delay after a server move. A name can keep the same spelling while its IP address changes. Some devices may still hold the earlier answer until it expires. One name can also lead to several servers that share the work.

Tell lookup from search and delivery

A search engine finds pages from keywords. DNS needs a particular host name and finds address information. A web server supplies the page itself.

If DNS fails, an internet link can still carry data to known IP addresses while name-based requests fail. "Server not found" can also follow a mistyped host name, so check the spelling before changing network settings.

A browser asks for a page and another computer sends a reply. This is the client–server pattern: a client requests a service and a server provides it. The same pattern supports files, games, messages and streaming.

A server is a role

A client can be a device or a program. Your browser acts as a client when it requests a page. A server can also mean a computer or the program answering requests. "Server" names the job, so an ordinary computer running suitable software can act as one.

Large services use powerful computers that stay on all day. Web servers supply web content; file servers share files; game servers coordinate online games; DNS servers answer name lookups. These roles provide different services over networks.

A website's pages, images and other files are stored on servers. When requested, copies travel to the client in packets. Viewing an image on a phone does not move the server's only copy onto that phone.

Streaming and downloading

A stream sends small parts of audio or video while playback continues. It can start playing before the whole recording arrives, but it needs a steady flow of data. If the connection falls behind playback, the player pauses to collect more data, called buffering.

A download saves a copy on the device. Some players can begin during the download. Once a usable file is fully downloaded, it can be opened without fetching more data from the server.

A stream using 5 Mbps for 30 minutes runs for 1,800 seconds:

$5\times 1800=9000$ megabits.

$9000÷8=1125$ MB.

That is a little over 1 GB of data. Streaming uses network capacity even though the learner never presses a button labeled "download".

Servers deliver messages and store files

Many messaging apps pass messages through their company's servers. A server can hold a message while the receiver's phone is off, then deliver it after that phone reconnects. The two phones do not always need to be online together.

Cloud storage keeps files on physical servers reached over the internet. An account controls entry. File permissions control who can view or change particular files. A cloud copy can protect work if a phone is lost; files stored only there need an internet connection to open.

Check how a service handles backups. A stored copy helps with device loss, while a separate backup that preserves earlier data helps after an accidental deletion. Copies that simply sync changes can sync the deletion too.

Data centers support large services

A data center houses many servers with cooling, backup power and fast network links. A service can keep copies in several centers. Nearby copies reduce travel delay, while another working copy can help when one center fails.

Servers also have capacity limits. A ticket website can become slow when many people request tickets together. Other websites may still work because the problem is the busy service, not every visitor's connection.

Using the internet does not always mean opening a web page. The internet is the worldwide network of networks. The World Wide Web is a collection of linked pages and files delivered across it.

Infrastructure and services

The internet brings together cables, radio links, routers, servers and shared protocols. Many providers, companies and governments run parts of it. No single company owns the whole network.

The web runs on this infrastructure. Browsers request resources using HTTP or HTTPS; URLs identify them, and HTML describes web pages. Hyperlinks hold URLs that lead to other resources, sometimes on servers in another country.

Picture a road network carrying several delivery services. The internet supplies the roads and connections. The web is one service using them. Email, messaging, video calls, games and smart devices also use internet connections.

The same service can be reached in different ways. Email in a browser involves web pages, while a separate email program can exchange email without presenting a website.

Browser, search engine and website

A browser is a program for viewing web pages. Chrome, Safari, Firefox and Edge are examples. A search engine, such as Google or Bing, is a service that finds pages from the words you enter. A website is a group of pages associated with a domain.

Opening a browser does not require a search. Entering a full URL requests that resource directly. Typing a topic into a search engine asks it to find possible resources. Search results then contain links that the browser can follow.

The internet came before the web

ARPANET, an early network that helped lead to the internet, sent its first message in October 1969. It switched to TCP/IP on January 1, 1983. Tim Berners-Lee proposed the web at CERN in 1989. These dates mark different developments, not two names for one invention.

Use the distinction to investigate faults

If one website fails but a game and email still work, the connection carries those services successfully. Check the site's typed address, its name lookup, the route to it and the server. One site's failure leaves several possible causes.

If all internet services stop on every device, examine shared parts such as the router and the provider connection. Local file sharing may continue because its data stays inside the LAN.

Opening a web page combines several jobs: find an address, send a request, carry its packets and process the reply. Following those jobs in order helps explain both a working page and an error message.

Find the destination

A student enters https://www.example.org/news/today.html on a school laptop. The browser separates the host www.example.org from the path /news/today.html. It needs the host's IP address before it can contact the server.

If a usable DNS answer is cached, the device can reuse it. Otherwise, a resolver helps obtain the address. In a worked diagram, let the returned address be 203.0.113.10, a reserved documentation value.

DNS supplies address information. The requested page will come from the web server.

Send the request across networks

HTTP gives the rules for requests and responses. HTTPS protects that exchange with encryption. The browser prepares a request for the page, and the device sends its data in packets toward the server's address.

The path from a wireless school laptop is:

1. Radio signals reach the access point.
2. Cable carries the data through a switch.
3. The school's router forwards the packets out of the LAN.
4. The provider and other networks' routers pass them onward.
5. The destination server receives the request data.

Each router chooses a next hop using destination information. Much of the journey is usually over fiber, even though it began on Wi-Fi.

Receive and display the response

The web server finds the requested resource and returns response data in packets. At the school's edge, the router directs replies to the requesting laptop. The return route can differ from the outward route.

For a transfer using TCP, missing data is resent and the data is delivered in order. The browser then interprets the received page. It often sends further requests for images and other resources named by that page. Some resources can come from other servers.

What appears on the screen is a locally displayed copy of data sent back. The browser does not reach into the remote computer's screen.

Read what an error tells you

A server's HTTP response starts with a status code. 200 means the request succeeded. 404 means the requested resource was not found. A 404 therefore shows that a server answered; reconnecting Wi-Fi does not repair a missing page path.

Error messages are clues to the stage reached. A timeout leaves more possibilities than a 404 because there was no timely response to interpret. Check what still works before replacing equipment or changing several settings.

Planning starts with people and tasks. List the devices, decide which ones move, and check how much data users need. Then choose an internet service and local equipment that support those needs.

Compare internet services

An internet service provider (ISP) connects a building or device to wider networks. It supplies the connection and address settings and usually a DNS service. Available technologies depend on the address.

Cost and coverage can keep people from reliable access. A fast plan unavailable in a remote area cannot meet that area's needs. Compare real choices at the location, not just the fastest advertisement.

Check both directions

Download speed is the rate for receiving data. Upload speed is the rate for sending it. A plan may offer much less upload capacity, which matters for sharing videos and video calls.

Uploading a 20 MB video at 10 Mbps takes at least:

$20\times 8=160$ megabits.

$160÷10=16$ seconds.

Use the upload rate for this task. A high download rate does not substitute for it.

Count every switch connection

A computer club has 4 desktops, 1 printer and 20 tablets. The desktops and printer stay in place, so cables suit them. The tablets use Wi-Fi through an access point.

The switch needs ports for:

• 4 desktops;
• 1 printer;
• 1 access point;
• 1 connection to the router.

That is 7 ports. An 8-port switch fits and leaves one spare. The tablets do not each need a switch port: they share the access point's wired connection. Forgetting its port or the router link makes a plan too small.

Draw, place and configure

Draw a labeled box for each device and lines for its connections. Mark cable and wireless links differently. A diagram can reveal a missing link before equipment is bought and help locate faults later.

Place the access point high and near the area's center, away from thick walls and large metal objects. Keep usual copper Ethernet runs within about 100 meters. More rooms or obstacles may need additional access points.

Choose a clear network name and a strong Wi-Fi password. Let the router assign addresses automatically to ordinary joining devices. Give shared equipment such as the printer a fixed address that does not conflict with other devices. After setup, test local sharing, wireless coverage and internet access from the places people will use them.

A useful fault report says what fails, where it fails and what still works. "One laptop cannot open any sites, but the others can" is more helpful than "the internet is broken." Start nearby and work outward, changing one thing at a time.

Check the simple causes

Check Wi-Fi is on and airplane mode is off. Check both cable ends and the link light. Try another device and another website. These comparisons separate a single-device problem from a shared one. Restarting a router can clear temporary faults, but test after the change so you know whether it helped.

Use network testing commands only on networks you have permission to test.

Read the settings

Open a command line window. On Windows, ipconfig shows the device's IP address and default gateway, the router used to leave the LAN.

An address of 192.168.1.20 and gateway 192.168.1.1 show configured settings. Test the link next; having the router's address written down does not establish that it is reachable. An IPv4 address beginning 169.254 can be a fallback after automatic assignment failed. Check the local connection and the service supplying addresses.

Test reachability and delay

ping sends small test packets and reports replies with round-trip times. ping 192.168.1.1 tests the local gateway. Then test an outside address and finally a host name.

Suppose 4 tests produce 3 replies, taking 21, 19 and 24 ms. One of the four is lost:

$1÷4\times 100=25\mathrm{\%}$ packet loss.

Average the replies that arrived:

$(21+19+24)÷3\approx 21$ ms.

A missing reply is not a zero-millisecond journey. Repeat measurements to understand a pattern of loss; weak signals, congestion and hardware faults are possible causes.

Use a test ladder

• If the gateway is silent, check the local link and whether it permits ping.
• If the gateway replies but outside addresses fail, investigate the router's internet link, ISP and possible filtering.
• If outside IP addresses answer but names fail, check DNS and spelling.
• If general access works and one site fails, check that site's address, lookup and service.

Some devices block ping. A silent server may still provide web pages, while a replying server may have a failed web service.

Look along the route

tracert on Windows and traceroute on macOS or Linux show responding hops toward a destination. Their timings can help locate delays. A * marks an unanswered test. If later hops respond, traffic continued past that point; the star alone does not locate a broken link.

Network data crosses equipment controlled by different people. Radio signals can also reach beyond a building. Encryption, access controls and filtering protect different parts of this journey.

Encryption protects readable content

Encryption transforms readable data into a form that needs the right key to recover it. A simple letter shift shows the idea: moving each letter three places turns "HELLO" into "KHOOR". Moving back three recovers the message. Real network encryption uses far stronger methods and long keys.

Wi-Fi security standards WPA3 and WPA2 encrypt the radio link. A password controls who can join a typical protected home network. An open network usually lacks this password-based protection.

HTTPS encrypts the exchange between a browser and a web server. Check the address and the browser's connection information; different browsers use different icons. HTTPS keeps people along the route from reading the protected content. A dishonest site can also use HTTPS, so encryption does not make its claims truthful.

Filter traffic and control access

A firewall checks traffic against rules and blocks what those rules reject. A typical home router allows replies to requests made from inside while blocking unexpected incoming connection attempts. The rules help limit access, but do not judge whether every page a person chooses to open is honest.

Accounts and permissions control which users can view or change shared files. Physical protection matters too: a locked equipment room makes it harder to reach servers, switches or cables directly.

Set up a safer shared network

Change the router's default admin password, which controls its settings. Choose a long Wi-Fi password for joining the network. These passwords serve different jobs. Turn on WPA3 or WPA2 and keep the router's software updated.

Give visitors a separate guest network so their devices stay apart from the main devices. Keep shared files behind suitable accounts and permissions. A guest who needs internet access does not need access to the school's file server.

Think about the whole route

A Wi-Fi password protects the radio link to the access point. The request then continues through other networks. HTTPS protects the browser–server exchange across that longer route. Both can help the same request without doing the same job.

On public Wi-Fi, check the network name with staff to avoid lookalike hotspots. Use HTTPS, and choose cellular service for banking or other private accounts. Entering information into a dishonest site still hands it to that site, even if the journey there was encrypted.