There are a number of different components in a typical enterprise computer network. Although most components are essentially just a computer, they can be categorised based upon their hardware and software capabilities and can be used in a network to serve a particular purpose.
End devices are those used by employees in a business or by other end users. They consists primarily of computers, including both laptops and desktops; servers, used indirectly by likely every client; and other consumer electronics such as smartphones and tablets.
A PC must meet a number of requirements in order to connect to a network and then send and receive data to and from other clients on the network.
In terms of hardware, a computer requires a network interface controller, or NIC. This is typically found in the form of a built-in motherboard network chip, PCIe network expansion card or wireless expansion card. A NIC is accompanied by a unique MAC address, which is used to identify the client on a network by switches and cannot be changed in the same way that an IP address can be. In addition to the MAC address given to a client by it's NIC, it is usually assigned an IP address in order for it to communicate across a network. This is configured by software, not hardware, however.
In terms of software, the client requires an operating system with networking drivers that allows for IP addresses to be configured. Popular choices for client operating systems include Microsoft Windows, OS X, Linux, iOS and Android. Support for DHCP can ease the process of building a network greatly, and this considered standard. IP address assignment can be achieved in one of two ways; static configuration or DHCP allocation. Static IP addresses only change when changed by an administrator, and must be specified on each machine individually. DHCP, or dynamic host configuration protocol, allows for a server to allocate machines an IP address within a certain range. This address can be used until the lease for it expires, which is generally around eight days, but clients should attempt to retrieve a new address when the lease is half-expired. The DHCP server is found by a client by way of a DHCPDISCOVER packet broadcast on the network.
PCs will be used in my network design as the terminals used by employees of the organisation. Laptops, tablets and smartphone will also be usable with the inclusion of a wireless access point.
Server hardware is effectively the same from a networking standpoint, although servers are often equipped with high-speed network interface devices. An example of this would be the inclusion of a fibre networking card to make the connection to switches, routers and other servers faster. A server may also include more than one interface to allow for redundancy or connection to multiple networks.
Servers require an operating system in the same way that clients do, although different software choices are generally made on the server-side. For example, OS X Server is available from Apple, although it sees minimal usage. Linux distributions such as CentOS, RHEL and Debian largely dominate the server market, with many businesses opting to use Windows servers for email and Office applications.
Servers will be used later in my network design to offer a number of services to the other end-devices. File sharing with FTP, IP allocation with DHCP and web hosting with HTTP(S) will all be available.
In order to allow multiple client devices to connect in a single network, and to allow several networks to communicate, interconnection devices are used. These account for most nodes within a network, when excluding clients and servers.
In order for a group of more than two computers to all be able to communicate with one another, a switch is required. All of the devices are connected to the switch either by wire or wirelessly with a WAP, and send information to the switch. The switch then analyses the packets of data sent to it and retrieves the MAC address of the destination. Using a table of physical ports and MAC addresses, the switch then forwards the package to the correct client.
In my network I will use switches to connect multiple devices. The reasoning behind this is outlined below; the alternative, a hub, is outdated and has several disadvantages.
Before switches were developed, a similar device called a hub was used. These are not as advanced as switches in that they do not target the information they transmit. When a packet is received from a connected device, the hub will place it in memory and then send a copy to every other device on the network. This means the the intended recipient will receive the packet, by un-intended recipients may too, making the use of encryption more desirable. It's worth noting that a switch, when experiencing high amounts of traffic, may revert to hub-like behaviour and start forwarding every packet to every client.
In addition to this, devices connected to a hub are in a single collision domain. This means that packet collisions are possible when multiple devices send data to the hub simultaneously. A switch keeps each device in a single collision domain, meaning that packet collisions when several devices transmit at once are theoretically impossible.
Hubs also only support half-duplex data transmission. This means that data may only travel in one direction along the cable at any one time. Full-duplex transmission, which uses the same cabling hardware, supports simultaneous movement of data in both directions. This is more efficient and allows for transfer of large files, for example, in both directions at once.
As stated, I will not be opting to use any hubs in the network I design. I have already decided to use switches, as they offer a number of advantages. Efficiency of operation, security, and data speeds are all improved when using a switch, and they have become the standard used in almost all networks.
In a similar way that a switch connects multiple end-devices together to form a LAN, routers are used to connect multiple LANs together as a WAN. A router is generally added to a network using a switch in the same way that an end device is, and is then connected to other routers. There are a number of tasks that are, or at least can be, undertaken by a router, other than the connection to other LANs:
NAT, or network address translation, means that routers wrap packets travelling from the LAN out into the WAN in another packet layer, also giving it an IP address to use on the WAN. When packets are sent back to a device on the LAN, the router unwraps the packet again and forwards it to the original client. In the network I design, NAT will be used if the network is connected to the internet or another WAN, but this is not planned.
Broadcasts are LAN-wide transmissions of data. If data is sent to the network's broadcast address, which can be found accurately by a client given the subnet mask, the switch will send the packet(s) to every connected device; one could image that a switch assumes hub-like behaviour. The package is read by every connected device, and certain services – including DHCP, explained below – may respond accordingly. Broadcast packets will be used indirectly in the network I design, as they will facilitate the assignment of IP addresses using DHCP.
Access control refers to the selective allowance of traffic through a router. For example, a router can be programmed to not allow certain devices to access the internet, forward traffic from some devices to a different server, and to block traffic to prohibited destinations. The time of day could also potentially be used to restrict access at certain times.
DHCP, or the dynamic host allocation protocol, allows clients to request an IP address to use on the LAN. This negates the need to manually configure the IP addresses on each device of a network. When a device first connects to a network, it will transmit a packet to a MAC broadcast address, which will ideally be replied to by a DHCP server. The response will include an IP address, lease time, and other DHCP data. DHCP will be used in the network design I create, as the time and hassle saved in assignment of IP addresses is very significant. DHCP is also mentioned in the PCs section.
QoS refers to the selective treatment of network traffic with intention of improving quality of service. An example of this is the prioritisation of VoIP call traffic to other networks over standard HTTP traffic. A second delay in a phone call could have a great impact on the effectiveness of the communication and could made the connection unusable. A second more of delay when loading a web-page or downloading a file would have a very minimal effect on productivity and would most likely go unnoticed. QoS is also used in some network cards marketed towards the computer gaming community. These network cards prioritise traffic to online game servers over other traffic, minimising the latency felt in-game.
I will use routers in my network to break the infrastructure up into more manageable sections. By separating the floors using routers, the amount of traffic filling the network in the form of broadcast packets will be reduced too. The routers will also allow for finer control over the network from the perspective of the system administrators, with each floor potentially being subject to different filtering configurations or service availability.
A wireless LAN, or WLAN, is a group of devices connected without wires, most commonly using the WiFi standards specified in the IEEE 802.11 documents. Wireless networks have seen a great growth in popularity in recent years, as more and more people have started using mobile devices as a key part of their lifestyles. Wireless networks are significantly more convenient to use, allowing a connected device to move freely within a certain distance of an access point.
As wireless protocols do not use a dedicated hardware link between end devices and the switch they use, security issues are raised. A wireless transmission can easily be listened to by any party within a given range, and the data analysed. If the data is not encrypted, personal and private technical information could both be left in the wrong hands. For this reason, several security protocols have been developed.
The oldest, WEP, has now been officially deprecated and should no longer be used. The protocol offers only weak encryption due to the use of a single or small number of private keys, and limits passwords to either 5 or 13 ASCII characters.
WPA was designed to success WEP and offer better security. The protocol introduced TKIP, which ensured that keys had not been tampered with, and offered further features such as user logins. One of the WPA protocol's weaknesses was its instruction to return to WEP security if WPA was not possible.
The greatest step forward was with the introduction of WPA2, which increased security even more. PSK, or pre-shared key, and AES encryption is more effective and suited to small networks, such as those found in homes.
A wireless access point will be used in the network I design in order to allow for the use of wireless mobile devices by employees. Standard security practices will be used, meaning that access to the network is safeguarded by a strong password and encryption is used to prevent the theft of data.
Most corporate networks make more heavy use of wired networking. This involves the use of physical copper or fibre lines to transmit data. Although use of physical wiring is more complex and difficult to manage, as well as more expensive to establish, the connections created are more reliable, more secure, and can carry data at greater speeds, which are all greatly important to many businesses.
Using copper wires at it's core, UTP cabling is cheap to manufacture and sees heavy use. Most commonly at the centre of a UTP cable are four pairs of wires, totalling eight lines for the entire assembly. Each pair of wires is twisted in the same direction, and the four pairs sometimes then twisted in the opposite direction. Each pair can often be found to be twisted at a different rate to the other pairs in the cable, a practice which is used to reduce electromagnetic interference.
A common use of UTP wiring can be found in Category 5, 5e and 6 cables. These use four twisted pairs, resulting in a total of eight wires. Most often found at each end of a Cat N cable is an RJ45, or 8P8C, connector, which allows a cable to be plugged into network devices and used to transmit data. Category cabling is very popular in computer networks due to it's relatively low cost and high data transfer speed (particularly with Cat 6).
UTP cabling will be used within my network design to connect client machines to their local switch. The need for fibre connection speeds at this level is not apparent, and use of such would be both costly and difficult to maintain. The use of Category 6 cabling means that data rates of a gigabit per second are easy to achieve; the connections are fast enough for a bottleneck not to be created.
Fibre cabling is an alternative to copper UTP, using glass fibres and beams of light to transmit data. The use of fibre rather than copper cabling is able to vastly increase data transmission speeds and has been used in recent years to improve home broadband connections.
Fibre cabling is usually found in a coaxial structure. Several layers of plastic, gel and foil may be wrapped around a cable, and more heavy-duty cables will also include layers to improve structural integrity. At the centre may be a single or cluster of glass fibres, each able to carry several signals. Fibre cabling can be categorised as either single-mode or multi-mode, referring to the structure of physical cable and the method by which light is sent into the class fibre. In a single-mode fibre, light is introduced with a laser, accurately and uniformly outputting light along the cable. The light is usually much brighter than that found in multi-mode cabling, and does not scatter or reflect nearly as much. Because the light travels closer to parallel to the cable, the core fibre can be much thinner compared with a multi-mode cable. Using LEDs rather than lasers, multi-mode cabling is not as accurate. A thicker glass fibre is used, and light is not introduced as uniformly meaning it frequently reflects on the inside of the fibre. Single-mode fibre allows for slightly improved transmission speeds, but is more expensive and time consuming to establish and maintain.
The light in fibre cabling is able to travel much faster than the electrons through a copper wire, which leads to vastly improved data speeds. Despite this advantage, fibre cabling is more expensive than UTP cabling, and is more delicate. Because of the numerous reflections within the fibre(s), the light signal deteriorates over distance as well, leading to the need for signal boosters at regular intervals along a cable. The tasks of splicing separate fibres together and terminating fibre lines are difficult and time consuming as well. In addition to this, glass fibres are very delicate, and can be broken easily. Despite all of these difficulties, fibre cabling is still used frequently. Fibre will be used in my network design for select parts of wiring, greatly improving the rate at which data can move around the network.
In order to demonstrate the use of network components without requiring real hardware, I used Cisco Packet Tracer to simulate a network. The brief given required a number of devices to be connected in a network, and the network needed to meet certain criteria. A download link to the Packet Tracer file (Packet Tracer 6.2+) is provided below.