Networking Basics, Explained in Plain English

103 concepts · 7 comparisons · 103 updated this month

Networking basics without the fluff: addressing, routing, switching, VPNs and ports, each with a short answer, a real diagram and a comparison table.

RFC 791IPv4, the address you still use
/24the subnet in most offices
65,535ports per transport protocol
1500bytes, Ethernet MTU
In this guide

Networking basics, from the first packet to a working network

Networking basics are the small set of ideas that explain how any two devices exchange data: how a device is addressed, how data is split into packets, how those packets find a path, and how the receiving end puts them back together.

A home Wi-Fi network, an office LAN and the internet itself all run on the same fundamentals, which is why learning them once pays off everywhere.

This guide covers computer networking basics in the order the ideas build on each other, the same networking fundamentals a certification course starts with: what a computer network is, how it works, the types of networks, the devices in them, the OSI and TCP/IP models, IP addressing, the common protocols and ports, wired and wireless links, basic network security, and how to troubleshoot.

Each section links to a full page in this library when you want the detail.

DefinitionWhat is networking?

Computer networking is the practice of connecting two or more devices so they can communicate and share data and resources. A computer network can be two laptops on one cable or millions of networks joined together, which is what the internet is.

The devices, called nodes, can be computers, phones, printers, servers or sensors. The connections between them, called links, can be copper cable, optical fiber or radio. A network is nothing more than nodes, links, and an agreed set of rules for using them.

Those rules are protocols. A protocol defines how data is formatted, addressed, sent, received and acknowledged, so that equipment from different vendors can talk without knowing anything about each other. The internet works because every device on it speaks the same family of protocols, TCP/IP.

What a network is used for comes down to four things:

  • Sharing resources: files, printers, storage and one internet connection used by many devices.
  • Communication: email, voice, video calls and messaging all ride on the same network.
  • Access to services: web applications, databases and cloud platforms reached from anywhere.
  • Central management: updates, backups, monitoring and security policy applied to many machines at once.

MechanicsHow a computer network works

Data does not cross a network in one piece. The sender breaks it into small units called packets, each carrying a slice of the data plus a header that says where it came from and where it is going. Packets travel independently, may take different paths, and are reassembled in order at the destination.

Every exchange, from loading a web page to joining a video call, follows the same steps:

  1. Name to address. The device asks DNS to turn a name such as example.com into an IP address.
  2. Local or remote. The device compares that address with its own subnet. A local destination is reached directly; anything else is sent to the default gateway.
  3. Finding the hardware. On the local network, ARP maps the next hop's IP address to its MAC address so the frame can be delivered.
  4. Switching. A switch reads the destination MAC address and forwards the Ethernet frame out of the one port that leads to it.
  5. Routing. Each router along the way reads the destination IP address and forwards the packet one hop closer, using its routing table.
  6. Delivery to the right program. At the destination, the port number hands the data to the right application, and TCP puts the packets back in order and asks again for any that were lost.

Addressing, switching, routing and ports: every other topic in networking is a refinement of one of those four.

Data transmission: who receives it, and in which direction

Two properties of a data transmission come up again and again in computer networking. The first is who receives it:

  • Unicast: one sender to one receiver, which is nearly all traffic.
  • Broadcast: one sender to every device on the local network, used by ARP and DHCP to find things. Too much broadcast traffic is the reason large networks are split into smaller ones.
  • Multicast: one sender to a group that asked for it, used for video streams and routing protocol updates. IGMP is how devices join a group.

The second is direction. A half-duplex link lets one side transmit at a time, like a walkie-talkie; a full-duplex link carries data both ways at once. Switched Ethernet is full duplex, and Wi-Fi is half duplex, which is one reason a cable is faster than radio at the same rated speed.

Where many devices share one medium, a media access control method decides who may transmit next, and on a reliable connection, flow control stops a fast sender from overrunning a slow receiver, which is the job of the TCP window size.

ScopeTypes of networks

Computer networks are usually classified by the area they cover. The names describe scope, not technology, and a single company often runs several of them at once.

TypeCoversTypical example
PAN, personal area networkA few meters around one personA phone, a watch and earbuds over Bluetooth
LAN, local area networkOne room, floor or buildingAn office network on Ethernet switches
WLAN, wireless LANThe same area as a LAN, over radioOffice or home Wi-Fi
CAN, campus area networkSeveral buildings on one siteA university or hospital campus
MAN, metropolitan area networkA city or metro regionMetro Ethernet between a company's city offices
WAN, wide area networkCountries and continentsBranch offices joined over MPLS or SD-WAN; the internet is the largest WAN
VPN, virtual private networkA private network carried over a public oneA remote worker's encrypted tunnel to the office

A VPN is the odd one out: it is not a physical scope but an encrypted overlay on top of another network. The VPN technologies page compares the ways of building one, and IPsec VPN covers the most common site-to-site form.

HardwareNetwork devices and what each one does

A handful of device types do all the work in a computer network: some connect multiple devices to each other, some connect multiple networks, and some control what may pass. Home and small office equipment often combines several in one box, which is why the roles get confused.

DeviceWhat it doesWorks with
Network interface card (NIC)Connects one device to the network and carries its MAC addressFrames
SwitchConnects devices inside one network and forwards each frame only to the port that needs it, using its MAC address tableMAC addresses
RouterConnects different networks and chooses the path for each packetIP addresses
Wireless access pointLets Wi-Fi devices join the wired networkRadio and frames
ModemConverts between the provider's line (cable, DSL, fiber) and EthernetSignals
FirewallAllows or blocks traffic between networks according to rulesAddresses, ports, applications
Load balancerSpreads incoming connections across several serversConnections and requests
Hub, repeater, bridgeOlder devices that repeat or join segments; switches replaced themSignals and frames

The two pairs people mix up most have their own pages: modem vs router, and what a load balancer does that a router does not. In a data center the first switch a server meets is usually a top-of-rack switch.

LayoutNetwork topologies

A network topology is the shape of the connections between the devices in a computer network. The physical topology is how the cables run; the logical topology is how the data flows, and the two can differ. The logical vs physical network diagram page shows the difference on one network.

  • Star: every device connects to one central switch. It is the layout of nearly every modern LAN, because one failed cable affects one device.
  • Mesh: devices connect to many others, giving several paths. Used between routers, in WAN cores and in mesh Wi-Fi.
  • Ring: each device connects to two neighbors in a loop. Common in metro and industrial networks.
  • Bus: every device shares one cable. Historic, and the reason early Ethernet needed collision detection.
  • Hub and spoke: branch sites all connect to one central site, the usual shape of a small company WAN.
  • Hybrid: a mix, such as stars joined by a mesh, which is what most real networks are.

ModelsThe OSI model and TCP/IP

The OSI model splits networking into seven layers, each with one job and each relying only on the layer below it.

Nobody builds to it exactly, but everyone uses it as the shared vocabulary: "a layer 2 problem" means switching, "layer 3" means routing, "layer 7" means the application. The TCP/IP model, which the internet actually uses, folds the same work into four layers.

OSI layerJobExamplesTCP/IP layer
7 ApplicationWhat the user's program speaksHTTP, DNS, SMTPApplication
6 PresentationFormat and encryption of the dataTLS, character encoding
5 SessionOpening and keeping a conversationSession setup, RPC
4 TransportDelivery between programs, by portTCP, UDPTransport
3 NetworkAddressing and routing between networksIP, ICMP, routersInternet
2 Data linkDelivery on the local network, by MAC addressEthernet, Wi-Fi, switches, VLANsLink
1 PhysicalBits on the wire or in the airCables, fiber, radio, connectors

The practical use of the model is troubleshooting from the bottom up: a link light before an IP address, an IP address before a DNS name. The layer 1 page starts at the cable, and the IPv4 header and TCP header pages show layers 3 and 4 field by field.

AddressingIP addressing basics

Every device on a network needs an address. It actually has two: a MAC address burned into the network card, used on the local network, and an IP address assigned by the network, used to reach it from anywhere.

  • IPv4 address: 32 bits, written as four numbers such as 192.168.1.20. The supply ran out, which is why private ranges and NAT exist.
  • IPv6 address: 128 bits, written in hexadecimal, with enough addresses that NAT is unnecessary.
  • Subnet mask: marks which part of the address names the network and which part names the device. A /24, the mask 255.255.255.0, holds 254 usable addresses and is the default in most offices. See what a subnet is and CIDR notation.
  • Default gateway: the router a device sends traffic to when the destination is not on its own subnet.
  • DHCP: hands out the address, mask, gateway and DNS servers automatically, so nobody types them in. See what DHCP is.
  • DNS: turns names into addresses. When it breaks, "the internet is down" even though the network is fine; see DNS not resolving.
  • NAT: lets many private addresses share one public address. The private IPv4 ranges are 10.0.0.0/8, 172.16.0.0/12 and 192.168.0.0/16. See NAT masquerading.

An address that starts with 169.254 means the device asked DHCP for an address and got no answer, which the link-local addresses page explains.

ProtocolsNetwork protocols and ports

Two transport protocols carry almost everything. TCP sets up a connection, numbers every segment, and retransmits what is lost, so it is used where every byte matters.

UDP just sends, with no setup and no retransmission, so it is used where speed matters more than perfection, such as voice, video and DNS lookups. The TCP vs UDP page sets them side by side.

A port number, from 0 to 65,535, identifies the application on a device. The first 1,024 are the well-known ports, reserved for standard services:

ServicePortTransportUsed for
SSH22TCPSecure remote command line
DNS53UDP and TCPResolving names to addresses
DHCP67 and 68UDPHanding out IP settings
HTTP and HTTPS80 and 443TCPThe web, and most APIs
SNMP161UDPMonitoring network devices
LDAP389TCPDirectory lookups
SMB445TCPWindows file sharing
RDP3389TCPRemote desktop

Below the transport layer, ICMP carries error and diagnostic messages, which is what ping uses. Between routers, routing protocols such as OSPF and BGP exchange the routes that fill routing tables; static vs dynamic routing explains when a network needs them at all.

LinksWired and wireless networking

Ethernet, standardized as IEEE 802.3, is the wired standard. Copper twisted-pair cable runs up to 100 meters per link and also carries power to phones, cameras and access points through Power over Ethernet, described on the 802.3af page.

Optical fiber covers longer distances and higher speeds; single-mode vs multimode fiber explains which one a given run needs. The standard Ethernet frame carries a payload of up to 1,500 bytes, the MTU, and MTU and fragmentation covers what happens when a packet is larger.

Wi-Fi, standardized as IEEE 802.11, lets devices connect without a cable, replacing it with radio transmission in the 2.4 GHz, 5 GHz and 6 GHz bands.

Wireless is shared and half duplex: every device on a channel takes turns, so a wireless network slows down as it fills up in a way a switched wired network does not. Channel planning matters for that reason, and the 5 GHz channels page lists the ones that are usable.

  • Choose wired for anything that stays in one place: desktops, servers, printers, access points themselves.
  • Choose wireless for anything that moves, and plan it with enough access points that each one serves a modest number of devices.

SecurityNetwork security basics

A network that carries everything is also the path an attacker uses, so a few controls belong in even the smallest one:

  • A firewall at the edge that blocks inbound traffic by default and allows only what is needed. See what a firewall is.
  • Segmentation. Separate guests, servers, phones and cameras with VLANs, so one compromised device cannot reach everything.
  • Encryption in transit. HTTPS and SSH inside the network, and a VPN for remote access instead of exposing services to the internet. VPN vs proxy explains why a proxy is not a substitute.
  • Strong Wi-Fi authentication, WPA2 or WPA3, with a separate guest network.
  • Access control on the network devices themselves. Named accounts and central authentication for every switch, router and firewall, instead of one shared password. TACACS+ vs RADIUS compares the two protocols that do it.
  • Patching and logging. Network devices need firmware updates like any computer, and their logs belong on a central collector, which is what syslog is for.

The security library takes each of these further.

DiagnosticsBasic network troubleshooting

Most faults are found by checking the layers from the bottom up and stopping at the first one that fails:

  1. Link. Is the cable in, is the link light on, is the Wi-Fi associated? Rising CRC errors point at a bad cable or port.
  2. Address. Does the device have an IP address, mask and gateway? ipconfig and ip show them.
  3. Gateway. Can it ping the default gateway? If not, the problem is local.
  4. Path. Can it ping an outside address? Ping and traceroute show where the path stops.
  5. Name. Does the name resolve? If an address works and a name does not, it is DNS.
  6. Application. Is the port open and the service answering?

When the network is up but slow, separate the two possible causes with packet loss vs latency. The network troubleshooting checklist is the long form of this list, and reading a Wireshark capture is the tool for whatever the checklist cannot explain.

Reading orderNetworking for beginners: a learning path in order

For someone new to the field, these pages in this order cover the networking basics without gaps, and each one takes about ten minutes:

  1. What a node is, and what a local area network is.
  2. Subnets, CIDR and the default gateway: how addressing works.
  3. DHCP and DNS: how a device gets its settings and finds names.
  4. ARP and the MAC address table: how switching works.
  5. TCP vs UDP and port numbers: how applications share a link.
  6. VLANs and inter-VLAN routing: how one network becomes several.
  7. Routing, then OSPF and BGP.
  8. VPNs and QoS: connecting sites and protecting voice and video.

GlossaryNetworking terms worth knowing

The networking concepts above lean on a small vocabulary. These are the terms the rest of the networking basics assume you know:

  • Protocol: an agreed set of rules for communication between devices, such as IP, TCP or HTTP.
  • Bandwidth: the capacity of a link, in bits per second. Throughput is what you actually get.
  • Latency: the time a packet takes to cross the network, usually measured as a round trip in milliseconds.
  • Packet and frame: the unit of data at layer 3 and at layer 2. A packet travels inside a frame.
  • MAC address: the 48-bit hardware address of a network interface, used on the local network.
  • Broadcast domain: the set of devices that receive each other's broadcasts; one VLAN is one broadcast domain.
  • Gateway: the device that leads out of a network, normally the router.
  • Client and server: the device that asks and the device that answers. In a peer-to-peer network every device does both.
  • ISP: the internet service provider, whose network connects yours to the rest of the internet.

Every networking concept, by topic

103 pages

Wireless2 pages

Cabling and power1 pages

Diagnostics7 pages

Network fundamentals5 pages

Fundamentals9 pages

Network operations4 pages

Routing14 pages

Infrastructure8 pages

Addressing5 pages

DNS2 pages

Switching7 pages

Remote access8 pages

Design6 pages

Ports15 pages

Network design1 pages

Network services1 pages

Cabling1 pages

Protocols7 pages

How to read the networking silo

Networking basics without the fluff: addressing, routing, switching, VPNs and ports, each with a short answer, a real diagram and a comparison table.

  • Start with the concepts at the top, in order. They take about an hour together and everything else refers back to them.
  • Use the index by topic. Each group maps to a chapter of the common certification outlines.
  • Go straight to the troubleshooting group. Each page has a checklist at the top.
  • The comparison pages end with a decision chooser and link to the companies listed in the directory that deploy them.

Networking questions we get most

Short answers here, full pages one click away.

What are the basics of networking?

Four ideas: addressing, which says where a device is; switching, which moves data inside one network; routing, which moves it between networks; and ports, which hand it to the right application. IP addresses, subnets, DNS, DHCP, TCP and UDP are the working parts of those four ideas, and everything else in networking builds on them.

What is networking in simple terms?

Networking is connecting devices so they can exchange data and share resources such as files, printers and an internet connection. A network is made of devices, the links between them, and protocols, which are the agreed rules for how the data is sent and understood.

What are the main types of networks?

By the area they cover: a PAN around one person, a LAN in one building, a WLAN when that LAN is wireless, a CAN across a campus, a MAN across a city and a WAN across countries. A VPN is a private network carried over a public one, so it can sit on top of any of them.

What is the difference between a router and a switch?

A switch forwards frames inside one network using MAC addresses. A router forwards packets between networks using IP addresses. Most small office devices do both, which is why the distinction gets blurry.

What is the difference between a modem and a router?

A modem connects your site to the provider, converting the signal on the cable, DSL or fiber line into Ethernet. A router connects your own devices to each other and shares that one connection among them. Many providers supply both in a single box.

What is an IP address?

A number that identifies a device on a network so that data can be delivered to it. An IPv4 address is 32 bits, written like 192.168.1.20. An IPv6 address is 128 bits and written in hexadecimal. Most office networks still run IPv4 internally.

What is the OSI model used for?

It is the shared vocabulary of networking. It splits communication into seven layers, from the physical cable at layer 1 to the application at layer 7, so that a problem can be placed: a layer 2 problem is switching, a layer 3 problem is routing. Real networks run TCP/IP, which folds the same work into four layers.

What is the difference between TCP and UDP?

TCP sets up a connection, numbers the data and retransmits anything lost, so it is used for the web, email and file transfer. UDP sends without setup or retransmission, so it is used where speed matters more than perfection, such as voice, video and DNS lookups.

What do DNS and DHCP do?

DHCP gives a device its IP address, subnet mask, default gateway and DNS servers when it joins the network. DNS then turns names such as example.com into the IP addresses that the network actually uses. When either one fails, a healthy network looks broken.

Is a VLAN the same as a subnet?

No, but they are usually paired one to one. A VLAN is a Layer 2 broadcast domain on a switch; a subnet is a Layer 3 address range. Putting one subnet on each VLAN is the convention that keeps routing simple.

Why does every office use a /24 subnet?

Because 254 usable addresses fit almost any single office, the mask 255.255.255.0 is easy to read, and DHCP defaults to it. Larger sites split into several /24s per VLAN rather than one bigger block.

What ports do I need to open for a VPN?

IPsec uses UDP 500 and 4500 plus IP protocol 50. SSL VPN uses TCP 443. WireGuard defaults to UDP 51820. Each VPN page lists the exact rules.

How do I tell packet loss from latency?

Latency is the round-trip time on a successful ping; loss is the share of pings that never return. Voice tolerates 150 ms of latency but breaks at 1 percent loss.

What should I learn first in networking?

IP addressing and subnets, because every later topic assumes them. Then DHCP and DNS, then how a switch and a router forward traffic, then TCP, UDP and ports. The learning path in this guide lists the pages in that order.

Other silos

One packet a weekA short, illustrated explainer every Tuesday. No vendor pitches, unsubscribe in one click.