You are viewing a free preview of this lesson.
Subscribe to unlock all 10 lessons in this course and every other course on LearningBro.
A network's topology is the shape it makes — the pattern in which its nodes are wired together and the pattern in which data flows between them. Topology is not a cosmetic choice. It dictates how the network behaves when a cable is cut, how it copes when traffic surges, how much it costs to build, and how easily it can grow. Two networks with identical hardware can have wildly different reliability and performance simply because of how they are arranged.
This lesson examines the four topologies on the OCR specification — bus, star, mesh and ring — and the crucial distinction between a network's physical topology (where the cables actually run) and its logical topology (the path the data conceptually follows). For each arrangement you will see a diagram, the engineering reasons behind its strengths and weaknesses, and the kind of scenario it suits. The recurring theme is trade-off: there is no "best" topology, only the topology that best balances resilience, cost and performance for a given situation. Learning to argue that balance is exactly what the exam rewards.
This lesson covers the parts of OCR H446 section 1.3.3 dealing with network topologies, specifically:
The skill being assessed is reasoned comparison: examiners credit candidates who explain why a topology behaves as it does and tie that behaviour to a scenario's needs, not those who simply label a diagram.
A topology describes the arrangement of the nodes (computers, switches, printers) in a network and the links between them. It influences four properties at once:
Every topology is a particular compromise between these. A design that maximises resilience (mesh) sacrifices cost; one that minimises cost (bus) sacrifices resilience and performance. Keep these four axes in mind as a checklist when comparing.
This distinction is subtle and frequently examined, so pin it down precisely.
| Type | Definition |
|---|---|
| Physical topology | The literal layout — where the cables run and how devices are physically wired together |
| Logical topology | The path data follows through the network, which may differ from the physical wiring |
The classic illustration is early Ethernet using a hub. Physically, every device plugs into a central hub, so the cabling forms a star. But a hub simply broadcasts every incoming signal out of all its other ports, so logically the data behaves exactly as if every device shared one common cable — a bus. The same physical star can therefore run a bus logical topology. A modern switch, by contrast, forwards each frame only to its intended recipient, so a switched star is a star both physically and logically. The lesson: never assume the picture of the cables tells you how the data actually flows — you must consider the behaviour of the central device too.
Why does this distinction matter beyond exam pedantry? Because the logical topology determines the network's real-world behaviour — its collisions, its security exposure, its performance — while the physical topology determines its cost and how you maintain the cabling. Two networks wired identically (physical star) can perform completely differently depending on whether the centre is a hub (logical bus, shared and collision-prone) or a switch (logical star, segmented and fast). When you analyse or troubleshoot a network you must reason about both layers: the physical layer tells you which cable to check when a single machine drops off; the logical layer tells you why the whole segment slows down under load. A second example: a wireless network is physically a star (every device associates with one access point) but its shared radio medium makes it behave logically like a bus — all devices contend for the same airtime — which is precisely why Wi-Fi, like the old bus, needs a collision-avoidance access scheme.
In a bus topology every device taps into a single shared cable — the backbone or bus. A terminator at each end absorbs the signal so it does not reflect back and corrupt other transmissions. Only one device can transmit successfully at a time; if two transmit at once, their signals collide and both must back off and retry.
graph LR
T1[Terminator]
T2[Terminator]
A[PC A]
B[PC B]
C[PC C]
D[PC D]
T1 --- A --- B --- C --- D --- T2
| Advantage | Explanation |
|---|---|
| Low cost | Uses the least cabling of any topology and needs no central device |
| Simple to set up | Easy to install for a small, temporary network |
| No central device | No switch or hub to buy or fail |
| Disadvantage | Explanation |
|---|---|
| Backbone is a single point of failure | One break in the main cable splits or downs the whole network |
| Collisions | All devices share one medium, so transmissions collide and must be retried |
| Poor performance under load | The more devices, the more collisions, so throughput collapses as the network grows |
| Hard to troubleshoot | A fault anywhere on the backbone affects everyone, and locating it is difficult |
| Weak security | Every transmission passes every device, which can "see" all traffic |
| Limited scalability | Adding devices degrades performance for everyone |
Bus was common in early, cheap LANs but is effectively obsolete; its collision-prone shared medium and fragile backbone made it unsuitable as networks grew.
The fatal weakness of bus is worth understanding mechanically, because it explains the behaviour of several later technologies. The single cable is a shared medium: at any instant only one device can be transmitting a clean signal. If device A and device C both begin transmitting at almost the same moment, their electrical signals overlap on the cable and corrupt each other — a collision. Both transmissions are ruined and must be sent again. The protocol that manages this on classic Ethernet, CSMA/CD (Carrier Sense Multiple Access with Collision Detection), has each device listen before transmitting (carrier sense), transmit only if the line seems idle, and detect collisions if two devices start together — whereupon both stop, wait a random back-off interval, and retry. The crucial consequence is that as more devices join a bus, the probability of two wanting the medium simultaneously rises, collisions multiply, retransmissions pile up, and usable throughput collapses long before the raw cable speed is reached. This non-linear degradation under load — not merely the fragile backbone — is the deep reason bus does not scale, and it is exactly the problem a switch solves by giving every device its own private link.
In a star topology every device has its own dedicated cable to a central device — almost always a switch in modern networks. Devices never connect directly to each other; all traffic passes through the centre. Because a switch gives each port its own collision domain and forwards frames only to the intended port, a switched star avoids the collisions that cripple a bus.
graph TD
SW[Central switch]
A[PC A] --- SW
B[PC B] --- SW
C[PC C] --- SW
D[PC D] --- SW
E[Printer] --- SW
| Advantage | Explanation |
|---|---|
| Fault isolation | If one cable or device fails, only that node is lost — the rest carry on |
| Easy to add/remove devices | Just plug into or unplug from a spare switch port; no disruption to others |
| Good performance | Each device has a dedicated link to the switch, and a switch removes collisions |
| Easy to troubleshoot | A fault is confined to one cable/port, so it is quick to locate |
| Disadvantage | Explanation |
|---|---|
| Central device is a single point of failure | If the switch fails, the entire star goes down |
| More cabling | Every device needs its own run back to the centre |
| Cost of the central device | A switch (and its ports) must be bought and maintained |
| Limited by switch port count | Growth beyond the switch's ports needs more switches |
Star is the dominant topology for modern wired LANs precisely because its fault isolation, easy expansion and collision-free switched performance outweigh the single-point-of-failure risk — a risk that can be reduced by using a reliable, redundant switch.
The performance advantage of a star comes from the concept of a collision domain — the set of devices that share a medium and can therefore collide with one another. On a bus, every device is in one big collision domain, so they all compete. A switch places each port in its own collision domain: the link between a PC and its switch port is used by just those two endpoints, so there is nothing to collide with, and a full-duplex switched link can send and receive simultaneously at full speed. In effect the switch segments what would be one contended medium into many private ones and intelligently relays frames between them. This is why upgrading a bus (or a hub-based star) to a switched star can transform a congested network without changing a single cable run: the topology of the wiring is unchanged, but the logical behaviour — and therefore the performance — is completely different. It is the clearest demonstration that the device at the centre of a star defines how the network actually behaves.
In a mesh topology nodes connect to multiple other nodes, creating redundant paths: if one link fails, data simply takes another route. This makes mesh the most resilient topology, at the cost of large amounts of cabling.
Every node connects directly to every other node — the maximum redundancy.
graph TD
A[Node A] --- B[Node B]
A --- C[Node C]
A --- D[Node D]
B --- C
B --- D
C --- D
Only the important nodes are richly interconnected; less critical nodes have fewer links. This keeps most of the resilience while cutting cost, and is how real large networks — including the Internet backbone — are actually built. The art of partial-mesh design is deciding which nodes deserve extra links: the ones whose failure would isolate many others, or whose links carry the heaviest traffic, are meshed most heavily, while peripheral nodes are attached more sparingly.
The number of links in a full mesh grows quadratically with the number of nodes. Each of the n nodes connects to the other n - 1, but that counts every link from both ends, so we halve it:
Working a value through: for n = 6 nodes,
The table shows how quickly this escalates — the reason full mesh is impractical at scale.
Nodes (n) | Links (n(n-1)/2) |
|---|---|
| 4 | 6 |
| 5 | 10 |
| 6 | 15 |
| 10 | 45 |
| 20 | 190 |
| 50 | 1,225 |
Notice that doubling the nodes from 10 to 20 roughly quadruples the links (45 -> 190): this quadratic growth is exactly why you cannot fully mesh a large network.
| Advantage | Explanation |
|---|---|
| High redundancy | Multiple paths mean a failed link does not cut off communication |
| No single point of failure | The network keeps working when any one link or node dies |
| Good performance | Traffic spreads across many paths, avoiding bottlenecks |
| Self-healing routing | Routers can detect a broken link and reroute automatically |
| Disadvantage | Explanation |
|---|---|
| Very expensive | Full mesh needs a huge amount of cabling and many ports per node |
| Complex to manage | Routing and configuration grow complicated as links multiply |
| Impractical at scale | The n(n-1)/2 link count makes full mesh unworkable for large n |
Mesh suits situations where reliability is paramount and cost is secondary: the Internet backbone, data-centre interconnects, and increasingly wireless mesh networks where nodes relay for one another without expensive cabling.
Subscribe to continue reading
Get full access to this lesson and all 10 lessons in this course.