Topologies

A topology is the arrangement of a network's nodes (devices) and links (connections). The physical topology is how the cables, switches and devices are really laid out. The logical topology is how data actually travels, which can be quite different: a Wi-Fi network has no visible cables at all but still has a logical structure.

The syllabus asks about three layouts, and for each you should be able to judge five things: reliability, transmission speed, scalability, cost and the chance of collisions (two devices sending at once).

Every device has its own cable to a central switch. This is how almost every home and office LAN is built. Reliability: good, because one broken cable only cuts off that device, though if the switch itself fails everything stops (a single point of failure). Speed: fast, because each device has a dedicated link. Scalability: easy, just plug into a free port, until the ports run out. Collisions: rare with a modern switch. Cost: more cable than some layouts, plus the switch.

In a mesh each node connects to several others. In a full mesh every node connects to every other, which needs n(n − 1) ÷ 2 links: 5 nodes need 10, 10 nodes need 45, 50 nodes need 1,225. A partial mesh links only some pairs. Reliability: excellent: if a link or node fails, data takes another route, and there is no central device to lose. Speed: good, but depends on the number of hops and the traffic on busy paths. Scalability: a new node must be linked to others, which gets costly in a full mesh (wireless mesh systems add nodes automatically). Cost: high, from the cabling and equipment. Used for critical infrastructure, and in wireless mesh Wi-Fi.

A hybrid topology combines others to get the best of each. A typical one is a mesh between a few important switches, with each switch the centre of a star for the devices near it. Reliability is high (several routes in the core, and an edge failure stays local), it scales by adding spokes or core nodes, but it takes more design, smarter routing and more management, so cost depends on how complex it gets. Large companies, campuses and telecoms networks use it.

The best way to understand reliability is to cause a failure. Pick a topology, then click a device or a link to break it. The page counts how many pairs of devices can still talk, and tests every possible single failure to find the worst one. The ring and bus are older layouts, included so you can see why they fell out of favour: a ring survives one cut but not two, and a bus is split by one break in its backbone.