The two simplest types of network topologies, or structures, in telecommunications networks are point-to-point lines and multi-drop lines. When point-to-point lines are used, each terminal is connected by its own line to a computer system. When multi-drop lines are used, several terminals share each data communications line to a computer. Obviously, point-to-point lines are more expensive than multi-drop lines: all of the communications capacity and equipment of a communications line is being used by a single terminal. Therefore, point-to-point lines are used only if there will be continuous communications between a computer and a terminal or other computer system. A multi-drop line decreases communications costs, because each line is shared by many terminals.
A star network ties end user computers to a central computer. In a ring network, local computer processors are tied together in a ring on a more equal basis. A bus network is a network in which local processors share the same bus, or communications channel. In many cases, star networks take the form of hierarchical networks. In hierarchical networks, a large headquarters computer at the top of the company’s hierarchy is connected to medium-size computers at the divisional level, which are connected to small computers at the departmental or work group level. A variation of the ring network is the mesh network. This uses direct communications lines to connect some or all of the computers in the ring to each other. Another variation is the tree network, which joins several bus networks together.
In most cases, distributed processing networks use a combination of star, ring, and bus approaches. Obviously, the star network is more centralized, while ring and bus networks have a more decentralized approach. However, this is not always the case. For example, the central computer in a star configuration may be acting only as a switch, or message-switching computer, that handles the telecommunications between autonomous local computers.
Star, ring, and bus networks differ in their performances, re-liabilities, and costs. A pure star network is considered less reliable than a ring network, since the other computers in the star are heavily dependent on the central host computer. If it fails, there is no backup processing and communications capability, and the local computers will be cut off from the corporate headquarters and from each other. Therefore, it is essential that the host computer be highly reliable. Having some type of multiprocessor architecture to provide a backup capability is a common solution. Star network variations are common because they can support the chain-of-command and hierarchical structures of most organizations. Ring and bus networks are most common in local area networks. Ring networks are considered more reliable and less costly for the type of communications in such networks. If one computer in the ring goes down, the other computers can continue to process their own work as well as to communicate with each other.