In much the same way as topography describes the lay of the land for terrestrial environments, a topology is used to describe the layout of a computer network: Which components are physically or virtually connected to which, how those connections are made or preserved, and so on.
Given the number of different requirements that networks are set up to meet, it should come as little surprise to learn that there are several distinct network topologies (the plural form of topology) in common usage – each with their own characteristics, and particular advantages or disadvantages.
In this article, we’ll be taking a closer look at network topology and the types most widely used in today’s IT environments.
How Network Topology Is Determined
A network topology describes the way that it’s arranged, including all of its nodes or intersecting points, and the lines connecting the various network elements. Topologies are typically illustrated in schematic or diagrammatic form, with symbols or icons representing the nodes, and lines depicting the connections or runs of cable.
The transmission/communication type and the protocols used in making connections may be described by a signal or logical network topology. The actual geometric configuration of workstations and cables is described by a physical network topology. Physical network topologies come in a variety of forms, as described in the following sections.
Bus Network Topology
A bus network topology relies on a common foundation (which may take the form of a main cable or backbone for the system) to connect all devices on the network. This main cable or bus forms a common medium of communication which any device may tap into or attach itself to via an interface connector. If only two endpoints form a network by connecting to a single cable, this is known as a linear bus topology.
All devices on the network are effectively connected to each other, so any communication sent onto the bus by a device is visible to all the other devices – but only the specific device for which the message is intended should access and process it. Data is typically transmitted in only one direction.
Bus network topologies based on Ethernet cabling are relatively easy and economical to install, though runs are limited by the maximum available length of cable. Expansion may be achieved by joining two bus cables together, but this topology works best with a limited number of devices (typically 12 devices or less, on a single bus). The system relies on its main cable for stability – and if this goes down, so does the entire network.
Star Network Topology
The star network topology has every computer or device on the network connected to a central server or hub, through which each workstation is indirectly connected to all the others.
Star network topologies are common in home networks, where the central connection point may be a router, switch, or network hub. Unshielded Twisted Pair (UTP) Ethernet cabling is typically used to connect devices to the hub, though coaxial cable or optical fiber may also be employed. Compared to the bus topology, a star network usually requires more cabling.
If a single node or workstation in a star network topology goes down, the other nodes remain unaffected. However, a failure of the central hub can take down the entire system. Fast performance is usually assured with few nodes and low network traffic, and the hub acts as a repeater for data flow in more complex environments.
Setting up the network, modifying it, and troubleshooting problems are usually straightforward affairs.
Ring Network Topology
Workstations and other network devices are connected in a closed loop, for the ring network topology. Within a single loop, data flows in one direction, with adjacent pairs of workstations being directly connected to each other. Indirect connections exist to more remote workstations on the ring, as information passes through one or more intermediate nodes.
Ring network topologies are most often found on school campuses, though some commercial organizations also use them. FDDI, SONET, or Token Ring technology are typically used. Data is transported bit by bit from each node until it reaches its destination. With large numbers of nodes, repeaters must be used to keep data signals “fresh” as they travel across the network.
In what’s known as a Dual Ring Topology, two connections may be established between each network node, allowing data to flow in two directions (clockwise and counter-clockwise).
As only nodes having “tokens” are permitted to transmit data, ring network topologies aren’t adversely affected by adding more nodes, or by high traffic conditions. Though cheap to install and expand, ring topologies are difficult to troubleshoot if problems occur. Notably, the failure of one computer can disrupt the entire network.
Mesh Network Topology
Communications made on a mesh network topology may take any of several possible paths from their source to their destination. In a full mesh network, each workstation or device is connected directly to every other. In a partial mesh network topology, some devices may be connected to all the others, while the remainder share information directly only with certain other workstations that they require priority communications with.
Mesh network topologies are typical of the internet, and certain wide area networks (WANs). Data may be transmitted via a routing logic, which is determined by set criteria such as “path of shortest distance” or “avoid broken links”. Alternatively, a technique known as flooding may be used, where the same information stream is transmitted to all network nodes, without prejudice.
A mesh network is typically robust, with each connection capable of carrying its own data load and provisions for security or privacy. Faults in a mesh topology may be readily diagnosed, but the cost of cabling is notably high (with bulk wiring required), while installation and configuration are usually quite complex.
Tree Network Topology
A tree network topology consists of two or more star networks, connected together. This is typically achieved by connecting the central servers or computers of the component star networks to a common bus, or main cable. So in effect, a tree network is a bus network of star networks. In the simplest form of tree network, only hub devices are connected directly to the tree bus, with each hub functioning as the root of a tree of devices.
Tree network topologies have a root node, to which all other nodes are connected in a hierarchy. For this reason, tree networks are also known as having a hierarchical topology. Typically, this hierarchy should have at least three levels.
Often used in wide area networks, tree network topologies are ideal for workstations which are situated in groups. The expansion of nodes and extension of bus and star topologies are easily accomplished and maintained. Error detection is also straightforward, but the systems tend to be cable-intensive and costly to install.
Maintenance of a tree network may become difficult as more nodes are added, and if the central hub fails, so does the network.
Hybrid Network Topology
The tree network topology described above is a special hybrid form extending the capabilities of bus and star-configured systems. It’s typical of an approach that combines two or more network topologies in order to maximize on the better qualities of the component parts.
Hybrid network topologies may be adopted (for example) to ease the burden of error detection and troubleshooting, or to allow a network to be scaled in size more easily. They can however lead to undue complexity and increased costs.
Physical vs. Signal Topology
Finally, a distinction should be made between physical topologies and the logical or signal topology governing the transfer of data.
In many cases, both physical and signal topologies are the same – but this isn’t always true. So for example, some networks may have a star network topology as they’re physically laid out, but data may be routed through them on a bus or ring network topology basis.
Share this Post