Networking - Hardware - Topology

Network topology refers to the physical layout of the devices in a network. Networks can have varying arrangements which can influence the way devices (also know as nodes) on the network talk to each other and transfer data. Sometimes the physical topology will vary from the logical operation of the network.

Different topology strategies are largely developed by the companies or industry groups who design a certain type of physical network hardware and the protocols used in conjunction with that hardware. The main goal is usually to allow maximum traffic flow thorugh the network. The implemented topology in any given network will vary based on the needs of that particular network, there is no universally correct physical layout. The needs of a small company with a small number of users and a single fileserver are going to vary quite drasticly from an large enterprise with multiple locations, departments and possibly thousands of network nodes. The most commonly used templates for network design are the bus, star, and ring layouts. Often times, complex networks will employ more than one of these topologies for different applications, thus creating a hybrid network.

Bus Topology
A bus orientation for a network consists of a sequential connection of nodes along a single cable. A bus network never loops and always has two distinct ends. Devices are connected to the bus in series. In this kind of network layout, data is transmitted down the line and individual nodes are expected to recognize their own addresses in the packets being transmitted. Nodes only receive the packets that are directed to their address. When receiving signals on the bus, devices will ignore packets with addresses belonging to other machines. Termination is required to keep signals from bouncing back along the cable causing interference and errors. A terminator basically consists of an resistor that absorbs and disapates excess energy and keeps it from being reintroduced into the network. The "daisy chain" resemblance of the bus topology makes it easy to configure, easy to add devices and very reliable because no central controlling hardware is required for operation. Some of the features that make a bus network so easy to use can also be its downfall. Without any sort of centralized switching, the security of such a network can easily be compromised and a single break anywhere along the bus can cause all nodes to lose communication.

Star Topology
The star topology network is easy to diagram because it resembles a star in logical representation. Cables radiate outwards from a central device with a node on the end of each cable. The central device will in most cases be a hub or a switch. By using a switch the security of the network is improved because traffice is only sent to the destination segment rather than broadcast to all segments. The main advantages of implementing a star topology are the reliability and ease of troubleshooting. If a single cable is faulty it is easy to isolate because only one node will be affected, leaving all other nodes to function normally without interruption. A downside to a star network cab be the cost. Wiring an office for this type of network can be high for a smaller business and there is also the cost of the central hub or switch to consider when deciding to implement a star network.

Ring Topology
A network laid out with a ring topology has all nodes connected ot each other in a closed loop. It is common for ring networks to use a token passing method of access control. With this method, a special piece of data called a token is passed from one node to the next. While a node is in possesion of the token, it has sole rights to transmit data on the network without concern of data collisions. To accept traffic from the network, a device must recognize traffic destined for its address and accept it, or pass it on to the next node in the ring. IBM's classic Token Ring networks had a hub device called a Multistation Access Unit (MAU). Each node had 2 connections to the MAU which made the physical configuration appear more like a star, but logically, it was still a ring. The MAU had special circuitry that allowed the ring to continue to function in the event that a device in the ring failed. This central piece of hardware enabled network administrators to centrally montior and reconfigure the network. Due to the collision free design, ring networks are often implemented in time critical applications such as process control and medical monitoring. The main drawback to this kind of network is the wiring cost and central hardware cost.