Networking

Local area networks are normally confined to a single building (hence the term "local"). Essential components of a LAN are:

• Servers
  Servers are the machines that centralise data, management, and security, amongst other things.
• Workstations
  Workstations are the machines upon which network users do their work.
• Network operating system (NOS)
  These are operating systems specifically designed for transmitting data over a network. Linux is the perfect example of such an NOS.
• Communication links
  Software and hardware designed to facilitate data transport.

However, LANs are not limited to these four components but support many other devices including printers, modems, nodes, etc.

The most common network types in use today are peer-to-peer networks, and client/server networks. These are examined below.

Peer to peer networks

Peer to peer networks are typically small (ten workstations). THey have no centralised server and, therfore, no centralised control but the ability to share files, peripherals and other resources. Each workstation has roughly the same capabilities as its peers. All machines can share their files and directories and all workstations can share the same printer.

The advantages of this type of network are simplicity and economy. The disadvantages are limited resource potential and a lack of centralised control.

Client/server networks

In client/server networks, workstations (clients) rely on data and services from one or more centralised servers. Advantages in this case are twofold: firstly an easier and more efficient management is possible through centralisation. By enforcing centralised file storage, it is possible to protect the system against even the most disorganised operator. Secondly, increased security is possible, as client/server environments allow for discretionary access control, allowing you to incisively grant or deny access to files, directories, and resources based on the user, time, date, etc.

A disadvantage is that clients rely on the server for data. If the server dies, the clients die with it. It is also more difficult to configue and administer this type of network, and it is inherently more expensive.

Once a network type is chosen, you must next choose a topology. Topology refers to the manner in which your network is wired - some types are examined below. The choice of network topology will be influenced by some considerations including:

• centralisation
• cost
• maintenance and troubleshooting
• scalability
• security
• speed
• stability
• distances
• single point failure:
  This is a physical or logical location (a server, hub, wire, etc) where one or more network devices are connected. When this connection fails, one or more workstations will be unable to transmit data. For critical tasks like transmitting DMX, the trick is to minimise the damage that occurs when that single point fails.

Bus Topology

The network is supported by a long, uninterrupted cable called a backbone. This backbone is the root of all connectivity. Network devices (workstations, perpipherals, ...) draw their network feed from the backbone. The backbone must be terminated at both ends.

Bus topology is simple, cheap and popular but difficult to troubleshoot. It lacks central administration and is subject to speed and performance constraints. Moreover, if the backbone fails, the entire network is knocked out.

Star topology

Star topology is significantly more structured than bus topology and focuses on centralisation. In a typical star network, each computer or peripheral is connected to a central point (usually a hub). Thus, failure of one connection will not usually affect the others.

A hub is a hardware device that centralises network activity. Network cables run from workstations to the hub. The hub then repeats hte signals it receives, and these are routed out again to other network devices.
Star networks offer centralisation, are usually faster, more stable and more easily managerd. This topology allows you to break the network into segments.

The main disadvantage of a star network is that it has a single point failure at the hub level. If the hub fails, all workstations will lose connectivity. Moreover, the star topology is more expensive due to the price of the hubs.

Daisy chain topology

This topology is also sometimes called a star-bus. Hubs (which could contain star network segments) are linked in series.

Daisy chain topology is used to inexpensively transform small LANs into medium-sized ones. However, daisy chaining many hubs can significantly reduce the network performance.

Mesh topology

Mesh topology is highly advanced. Its purpose is to interconnect two or more sizeable LANs. Isolated networks are given multiple paths to their counterparts through routers. This way, total system failure is unlikely.

The main advantages are stability and safety, but it is expensive and requires a high level of knowledge.

Switched topology

Switched networks closely resemble star networks but work differently. Network devices are attached to a central switcher (smart hub). Bandwidth is used on a device-by-device basis: each device is allocated the full bandwidth, rather than sharing the available bandwidth between all devices.

Switched networks are far superior in this respect, but are also quite expensive.

Ethernet is a LAN technology (originally developed by Xerox) that connects computers and transmits data between them. Data is packaged into small units called frames or packets and sent over wires. Strictly speaking, Ethernet complies to the international standards ISO/IEC 8802-3 or IEEE 802.3

To describe an Ethernet LAN, we can use the analogy of a large pipe with outlets spaced along its length, used as a speaking tube by a number of people. Each person listens at an outlet for any message shouted down the tube that might be meant for him or her. If you have something to say, you listen until no-one else is speaking, and then you shout your message. The pipe can carry a lot of messages, because it's a big, fast pipe (Ethernet most commonly runs at 10 Mbits per second, 40 times faster than DMX). Furthermore, information can go back and forth between any two listeners (or nodes in network jargon) without having to go through some central authority.

Collisions occur on an Ethernet network because all computers broadcast without waiting for a turn, sometimes running messages into each other. Ethernet is designed to correct this, but if there are too many computers on the network, collisions can slow down the network performance.

There are several hardware areas that need to be covered in a network.

Cabling and connectors

Popular types of cable are: