A PAN, or a personal area network, is used from close range connections between devices. An example of a PAN network would be the Bluetooth connection established between a pair of wireless headphones and a smartphone the unit is pair with. A PAN network generally does not include more than two devices.
A LAN, or a local area network, refers to a network of several computers connected typically via a switch. Each of the connected computers are assigned IP addresses either dynamically with DHCP or statically during configuration. The members of a LAN are able to connect to one another, enabling applications such as file sharing, printer sharing and media streaming. The management of a LAN is usually handled by a system administrator, although smaller home networks can often work effectively using only automatically configuration.
A WLAN is a LAN using a wireless connection protocol such as IEEE 802.11 ac rather than a wired system such as Ethernet. WLANs have become particularly popular in recent years, commonly known as Wi-Fi, and are often found in homes, business offices and public areas including airport terminals and cafés.
A MAN, or a metropolitan area network, is a WAN spread over a particular area, often a city. The term has not seen wide usage, as a MAN is essentially a small WAN and is technologically indifferent. MANs are often referred to by ISPs when working to improve internet connection speeds, as fast and reliable connections between MANs improve consumer broadband speeds.
A WAN, or a wide area network, is a network of LANs. In a similar way that a LAN consists of several end-point machines, a WAN consists of several LANs. Each LAN is assigned a global IP address that can be used to access machines on one LAN from another, granted port forwarding has been configured. The largest example of a WAN is the internet. LANs in homes, offices, educational institutions and data centres are all connected in a single WAN, allowing them to all communicate with one another.
Wireless networks are those which use no wires, and exist at both small and large scale. For example, the 2G, 3G and 4G mobile communication networks span the majority of the globe, and provide telephone and data services to handsets within a huge area. This is thanks to the fact that a single telephone mast is able to provide coverage over as much as a 50 km-radius area, granted the signal is unobstructed. The standards – GSM, IMT-2000 and LTE respectively – dictate how data should be formatted by all compliant infrastructure; both hardware and software. This ensures maximum compatibility across nations and around the world. The protocol definitions for wireless communication are found under the IEEE 802.x series of standards, along with those for PANs and MANs.
As briefly mentioned before, there are software standards used to ensure that communication between hardware and software developed by different organisations is successful. Examples of protocol standards used in each of the network models are given below.
PANs most often use the Bluetooth standard, initially specified by the IEEE 802.15.1 documentation and now maintained by the Bluetooth Special Interest Group, or SIG. The protocol allows mobile devices and peripherals to communicate over short distances, and is typically used for low- to zero-configuration applications. Bluetooth enabled smartphones can share files when paired using the protocol, and it can also be used to connect wireless computer peripherals to a USB transceiver. Bluetooth uses low-power radio waves at a frequency of 2.4 GHz, and can typically transmit within a range of about 10m. Other standards designed to enable personal area networks are defined in the standards 802.1.1 to 802.1.6.
Wired LANs are most often built using the Ethernet protocol to connect several clients, servers, switches and routers together. This protocol is defined by the IEEE 802.3 working group. The protocol was designed to succeed other protocols such as token ring and and ARCNET, which are not able to support modern network bandwidth and speed requirements. The Ethernet protocol typically uses Category 5 or 6 cabling, commonly referred to as Cat 5 and Cat 6, with 8P8C (eight-position eight-contact) connectors at each end. These connectors are often referred to as RJ45 connectors, and the completed cable an Ethernet or patch cable. The Ethernet protocol covers software as well as hardware, specifying how data packets should be encoded and parsed. The Ethernet protocol aims to prevent packet collisions in a network using CSMA/CD, or Carrier Sense Multiple Access with collision Detection, which involves listening on a cable for existing connections before transmitting data. A 32-bit jamming signal can be transmitted by a network node to ask others to stop transmitting data.
Wireless LANs, or WLANs, most often use the Wi-Fi standard to transmit data. This protocol uses radio waves to send and receive data without the need for wires, and is most commonly used to create wireless internet access points in homes, workplaces and public areas. The Wi-Fi standard is defined in the IEEE 802.11 set of specifications, and is available in several generations. The different version of the protocol are identified with a suffix letter; either -1997, a, b, g, -2007, n, -2012 or ac. Other identifiers used include af, ai and ax, but these standards are not in common use. Wi-Fi signals are broadcast as radio waves, with an unobstructed range of around 100m, and usually assume a frequency of 2.4GHz or, more recently, 5GHz. 3.6GHz and 60GHz are also known to have been used. Many wireless hotspots are secured with WEP or WPA password-protection and encryption, the former of which is now a deprecated standard.
MANs, being technologically very similar to WANs, use the same protocols as these networks.
WANs have used a number of different protocols throughout time, with Point-to-Point Protocol over Ethernet, or PPPoE, currently the most common. This allows an Ethernet connection to be carried by a copper or fibre connection to an ISP's PPPoE server. The PPP protocol is standardised by the RFC 1661 document, and PPPoE by the RFCs 2516, 3817, 4638 and 4938. PPP is considered a layer 2 (data-link layer) protocol in the Open Systems Interconnection, or OSI, networking model. Before the PPP protocol was standardised and in common use, the Frame Relay protocol saw frequent use.
There are a series of protocols used within the TCP/IP protocols considered to be application-layer. All services utilising the web conform to the standards of one or more application layer protocols. It is possible to design one's own protocol for a particular application, and assign it a standard port number. The protocols used in the transmission of an email are explained below.
Firstly, the email will travel from the source client to an SMTP, or Simple Mail Transfer Protocol, server. Credentials are often required to use an SMTP server, and the servers are usually provided by mail services such as Gmail and Outlook.com. The SMTP server also does not have to be related to the sender's address; one could use Google SMPT servers to send mail from an email address on their own domain.
The message then travels through the internet to another SMTP server, identified as the server responsible for the email address. For instance, if an email was sent to
smtp-mail.outlook.com would be found. The email would be sent to this server, which would forward it to the next server via IMAP or POP standards.
An IMAP (Internet Message Access Protocol) or POP (Post-Office Protocol) server would receive the email, and place it in the correct inbox. These two protocols work in slightly different ways; although they can both be configured to minkc the other's behaviour, an IMAP server typically leaves mail on the mail server when using an email client such as Thunderbird or Outlook, whereas a POP server deletes its copy once the client has received its own. Most users prefer the behaviour of IMAP, as email messages, however old, are accessible from any computer with internet access. POP, alternatively, requires a user to have their home or work computer with them in order to access old emails. Once an email has arrived at either a POP or IMAP server, the recipient should be able to access it.
Connections within the modern internet take many different forms. The format in which data is encoded is typically defined by a standard. IMAP and POP are examples of application-layer protocols, the latest versions of which have been defined by RFCs 3501 and 1081 respectively.
A protocol works by defining a number of practices and formats of data that should be exercised and used on the web. By following common standards, hardware and software from numerous companies and developers can be assured to work together. For instance, a HTTP reply from a web server should begin with a series of headers informing the client about the connection. This includes information hostnames, cookies, content-types, hashing and more, and can be used to great effect in web-development. The key aspect of this, however, is that all browsers – Google Chrome, Mozilla Firefox, Microsoft Spartan, etc. – and web servers – Nginx, Apache, Hiawatha, etc. – follow the same rules. All client and server software packages accept the same parameters in HTTP headers, and return data using those parameters. The connections are established in the same way, and can offer features to a developer, granted support at both the client and server sides, such as persistent, Keep-Alive connections.
Protocol standards are written in order to finalise how computers should communicate for particular tasks. Standards in computing may seem a foreign concept to some, but standards and protocols are sued everywhere. When travelling on the roads in most countries, for instance, one must drive in the right-hand lane. Signs at the roadside and markings on the tarmac give drivers information on how they must behave in certain areas. The Highway Code is a form of protocol followed by drivers, and without it disagreements would arise and accidents inevitably happen.
By using protocols when transferring data across the internet, one can ensure that connections are successful, data is treated correctly and security is upheld.
Local area network models can be sub-categorised into one of the following groups based on their design.
A P2P (peer-to-peer) network, typically only useful at a small scale, involves each networked device being connected physically to a switch or wirelessly to an AP (access point). This allows all of the devices to communicate with one another, and can provide an internet connection if a router is connected to the switch as well. In a P2P network, each device is equal, and can act as a server to the other devices or as a client. For instance, a game server could run on one computer, and three other computers running the game client could connect to the server. This model is useful for small networks that need only offer minimal functionality, and is efficient once configured. However, when more than around eight devices must be networked together, difficulties arise. If users need to be able to use the same credentials to log in on several machines, other devices need to be freely added to and removed from the network, and users want the benefits brought by centralisation, the client–server model is more appropriate.
The client–server model, in contrast, has machines dedicated as servers and others as clients; a networked device will either listen for and respond to requests from other devices, or will create those requests based on user actions. Typically a physical server, or a number of them, are used to store files, login credentials and software, and to run applications such as web, mail and source-code management services. This allows, for instance, a user to us the same login credentials on any applicable device without their user profile being set up individually on each client. This is often combined with directory services, which allows users access to specific filesystem directories on the server side regardless of the client machine they're using. The use of a server makes the addition and removal of devices to and from the network much easier, with services like DHCP being used for automatic IP-address assignment becoming feasible. In a client–server network, clients are still able to act as servers, meaning functionality is not lost. Using this model is simply more efficient, easier to manage and more convenient for end users.
A P2P network is considerably easier to set up from scratch. All one needs is a switch, or more realistically a home broadband router/switch/access point, cabling and/or wireless devices, and client machines. The devices are connected to the central device, which acts as a switch. Many routers shipped by ISPs in fact perform several tasks, and one of these is typically the running of a DHCP server. One could argue that this is enough for a usual home network to class as a client–network, although most would not consider it such. Most P2P networks require IP addresses to be configured manually, which requires some skill but it more reliable in the long-term.
A client–server network is much more complex to establish. More hardware is required; more cabling, switches and importantly servers, are required. The network is centred around a server or multiple servers, and these must therefore be configured and connected to the switches in the network. As the network grows, the need for more capable servers does too, leading to further expense. That said, for a large-scale network, the client–server model is much more suited. Users need only be created on one machine before someone is able to authenticate on any connected client, and adding public network drives and private storage for specific users is much simpler.
In a large peer-to-peer network, it is very difficult to manage a number of users. For a small number of machines which are typically used by a single person each, the client–server model's directory services capabilities are not of much use, but P2P network designs are not suitable for organisations and educational institutions. A P2P configuration allows for more control over individual machines, which can have advantages. For example, most people do own dedicated server hardware. In order to host a server within their home LAN, they may choose to re-purpose an ageing computer, in which case several shared user credentials within a network would not need to be known by the all networked devices.
The client–server model has the upper hand for larger organisations, however. Installing software on a number of different computers is not easy, and this can be made easier with a centralised network drive. Software can be installed once on the server, making it easier to upgrade in the future, and client machines made aware of the installation. This saves space as well as man-hours in upgrading. The time saved when creating a new user to be accessible on several computers is also very significant. The credentials are stored on the server, meaning any machine connected to the network will accept new users instantly.
A peer-to-peer network is much harder to scale than a client–server network. Adding clients involves configuration of IP addresses on each individual device, and this is particularly inconvenient for public internet hotspots. A client–server network typically allows for the rapid addition and subtraction of nodes from the network if DHCP is in use, saving time for both technicians and end users.