Networks can be classified by their topology, which is the basic geometric arrangement of the network. Different types of network configurations exist for network designers to choose from. Communications channels can be connected in different arrangements using several different topologies. This arrangement allows users to exchange information and share resources (software and hardware).
Four basic types of network configurations are star, bus, ring, hierarchical and mesh. Ring, bus, and star topologies are commonly used in LANs and BNs. Star and mesh topologies are commonly used in MANs and WANS. The networks are usually built using a combination of several different topologies.
A star topology is one in which a central unit provides a link through which a group of smaller computers and devices is connected. The central computer is commonly called a host computer. A host computer is usually a large computer such as a minicomputer or a mainframe. A file server is a large storage device that provides volumes of data and programs to the other units in the network.
In the star network, all interactions between different computers in the network travel through the host computer. The central unit will poll each to decide whether a unit has a message to send. If so, the central computer will carry the message to the receiving computer.
Star networks represent a very popular form of configuration for time-sharing systems in which a central computer makes available resources and databases for several "client" computers to share. As such, the star network is appropriate for systems that demand centralized control. The disadvantage of the star network is that a processing problem in the central computer can be paralyzing to the entire system.
In a star network, the central unit may be a host computer or a file server. The host computer is a large centralized computer, usually a minicomputer or a mainframe. In contrast, the file server is a large-capacity hard-disk storage device. It stores data and programs files shared by the users on the network. Also, called a network server.
In a bus configuration, each computer in the network is responsible for carrying out its own communications without the aid of a central unit. A common communications cable (the bus) connects all of the computers in the network. As data travels along the path of the cable, each unit performs a query to determine if it is the intended recipient of the message. The bus network is less expensive than the star configuration and is thus widely in use for systems that connect only a few microcomputers and systems that do not emphasize the sharing of common resources.
The problem in a computer on a bus topology does not frustrate the operation of the network, but a crack in the central cable will stop the whole network. Bus topology is popular because many computers can be connected to a single central cable. In a bus topology, each end user computer in the network handles its own communications control. There is no host computer or file server. As the information passes along the bus, it is examined by each terminal to see if the data is for it.
A ring configuration features a network in which each computer is connected to the next two other computers in a closed loop. Like the bus network, no single central computer exists in the ring configuration. Messages are simply transferred from one computer to the next until they arrive at their intended destinations. Each computer on the ring topology has a particular address. As the messages pass around the ring, the computers validate the address. If the message is not addressed to it, the node transmits the message to the next computer on the ring.
This type of network is commonly used in systems that connect widely dispersed mainframe computers. A ring network allows organizations to engage in distributed data processing system in which computers can share certain resources with other units while maintaining control over their own processing functions. However, a failure in any of the linked computers can greatly affect the entire network.
The ring arrangement is the least frequently used with microcomputers. However, as stated above, it often is used to link mainframes over wide geographical areas to build distributed data processing system. The loss of a mainframe usually does not restrain the operation of the network, but a cable problem will stop the network altogether.
A hierarchical network (or a tree network) resembles a star network in that several computers are connected to a central host computer (usually a mainframe). However, these "client" computers also serve as host computers to next level units. Thus, the hierarchical network can theoretically be compared to a standard organizational chart or a large corporation. Typically, the host computer at the top of the hierarchy is a mainframe computer. Lower levels in the hierarchy could consist of minicomputers and microcomputers. It should be noted that a system can sometimes have characteristics of more than one of the above topologies.
This topology is effective in a centralized corporation. For example, different divisions within a corporation may have individual microcomputers connected to divisional minicomputers. The minicomputers in turn may be connected to the corporation's mainframe, which contains data and programs.
This is a net-like communications network in which there are at least two pathways to each node. In a mesh topology, computers are connected to each other by point-to-point circuits. In the topology, one or more computers usually become switching centers, interlinking computers with others.
Although a computer or cable is lost, if there are other possible routes through the network, the damage of one or several cables or computers may not have vital impact except the involved computers. However, if there are only few cables in the network, the loss of even one cable or device may damage the network seriously.