Learn
A2.2.1 | FUNCTIONS AND PRACTICAL APPLICATIONS OF NETWORK TOPOLOGIES
01 | HOW SHOULD THE DEVICES CONNECT?
A school opens a new computer room. The computers need to share resources, but the technician must decide how to connect them. Should every computer connect to a central device? Should important network devices have several routes between them? What happens when a cable breaks?
A network topology is the arrangement of network nodes and their links. A node is a participating device or connection point. A link is a connection between nodes. Topology describes their relationships, not where desks happen to stand in the room.
This lesson develops three arrangements: star, mesh and hybrid. We will connect each arrangement to reliability, transmission speed, scalability, data collisions and cost, then apply those factors to homes, small offices and larger organisations.
There is no topology that automatically wins every comparison. A sensible design meets the users’ needs within the available budget and support capacity.
02 | PHYSICAL AND LOGICAL ARRANGEMENT
Physical topology concerns actual connections, such as cables between computers and a switch. Logical topology concerns how communication flows or is organised over those connections. These descriptions can differ.
For example, an old Ethernet hub and a modern switch can both sit at the centre of a physical star. A hub repeats traffic onto a shared medium; a switch selectively forwards frames and can support separate full-duplex links. The shape alone does not tell you whether collisions occur.
When reading a diagram, ask what each node represents, which links exist and which paths are available. Lines crossing on a drawing are not necessarily joined. A path is a sequence of links connecting a source to a destination.
A topology is also different from a client-server or peer-to-peer model. Those models concern service roles; a physical star can support either arrangement.
03 | STAR: A CENTRAL CONNECTION POINT
In a star topology, each outer node connects to a central device. In a common wired LAN, that device is a switch. When computer A sends a frame to computer B, the frame passes through the switch rather than through the other computers.
The dedicated connections make faults easier to isolate. If A’s cable fails, A loses that connection while B, C and D can still communicate through the functioning switch. Disconnecting one computer does not normally require breaking the other computers’ links.
However, this simple star depends on the central switch. If the switch fails completely, the attached computers cannot communicate through it. That switch is a single point of failure: one failed component can interrupt the service for all dependent nodes.
04 | STAR IN A HOME OR SMALL OFFICE
Imagine a small design office with eight computers, a shared printer and a file server. A switched star provides a straightforward connection plan: connect each device to a suitable switch port, then connect the switch to the router for external network access.
Adding another wired computer is often simple if a spare port, cabling route and sufficient capacity are available. A larger switch or additional access switch may be needed when the original fills up. Scalability means more than finding a spare socket: uplink capacity, power and management also matter.
For an office that depends on its files, the central device’s quality, support arrangements and power protection deserve attention. A spare replacement switch can reduce repair time, but it does not provide automatic failover while sitting in a cupboard.
A home router often combines a switch, wireless access point and router in one case. That convenience does not remove dependencies: a failure of the combined unit can affect several services at once. Wireless devices also share radio resources, unlike separate wired full-duplex Ethernet links.
05 | MESH: PROVIDE ADDITIONAL PATHS
A mesh topology connects nodes with multiple interconnections. In a full mesh, every node has a direct link to every other node. In a partial mesh, only some of those possible direct links exist.
The main attraction is path redundancy. If a direct link between A and B fails, a suitable mesh may still provide a path from A through C to B. This can keep communication available while a fault is repaired.
The word “may” matters. The remaining links must actually connect the endpoints, and forwarding or routing mechanisms must be configured to use them. More lines on a diagram do not automatically provide working failover. A partial mesh can still contain vulnerable nodes or links.
Mesh nodes are often routers, switches or wireless infrastructure devices rather than every user’s laptop. A large organisation can add resilience to its backbone without buying a direct cable between every pair of desktop computers.
06 | FULL MESH: THE COST OF EVERY POSSIBLE LINK
Consider four nodes. Each needs three direct links, but counting 4 × 3 counts each shared link twice: A–B and B–A are the same two-way physical connection in this example. There are therefore 4 × 3 ÷ 2 = 6 links.
For n nodes, the number of point-to-point links in this full-mesh model is n(n − 1) ÷ 2. Each node needs n − 1 connections. Moving from five nodes to six adds five new links, not just one.
This explains a scalability trade-off. Full mesh provides many alternative paths, but ports, cabling, installation, configuration and maintenance grow rapidly. Partial mesh can focus extra links on the places where failure would matter most.
The calculation counts logical point-to-point connections in this model, not necessarily separate cables in a shared wireless system. Wireless mesh infrastructure has different physical and radio-resource constraints.
07 | HYBRID: COMBINE ARRANGEMENTS FOR DIFFERENT NEEDS
A hybrid topology combines different topology arrangements in one network. For example, a college can use local switched stars in classrooms and a partially meshed core connecting important network devices.
The local stars keep endpoint connections manageable. The interconnected core gives selected inter-building routes alternative paths. This balances cost and resilience instead of requiring a full mesh of every student device.
A hybrid inherits the strengths and weaknesses of its parts. A classroom computer with only one cable still loses access if that cable fails. Two core paths cannot repair that missing local connection. Likewise, a classroom switch without redundancy can still interrupt all of its attached computers.
Hybrid is not a single fixed shape or a promise of high reliability. You must describe which arrangements are combined and where the dependencies remain. Several stars connected as a simple hierarchy may be described as an extended star or tree; the star-plus-mesh example here makes the combination explicit.
08 | TEST A FAILURE AND TRACE THE REMAINING PATH
Try sending data between the named endpoints. First observe the normal route, then remove a link or node. A highlighted path shows reachability in the remaining graph. Grey dashed links and pink nodes identify failures.
This is a teaching model of connectivity, not a simulation of real switch protocols, radio performance or failover timing. It assumes working path selection and sufficient capacity. A route existing does not prove that a real network will recover instantly.
09 | RELIABILITY: FIND THE ACTUAL DEPENDENCIES
Reliability concerns dependable operation; availability concerns whether a service is accessible when needed. Redundant routes can improve resilience to certain faults, but a topology diagram alone cannot tell you an exact availability percentage.
Ask what must remain operational for a particular journey. A star’s single endpoint cable is essential for that endpoint, and its central switch is essential for communication through that star. In a mesh, removing one link may leave a route, while removing several related links may split the network.
Failures are not always independent. Two “redundant” fibre routes in the same underground duct can both be cut by the same excavation. Two switches on one failed power supply can both stop. Resilience requires attention to shared risks as well as extra links.
For a government office handling essential services, diverse routes, suitable power arrangements, monitoring and tested recovery can justify additional expense. For a small home network, a simpler design may meet the required level of service.
10 | TRANSMISSION SPEED: LINKS, LOAD AND BOTTLENECKS
Topology influences possible paths and where traffic gathers, but it does not set a universal speed. Link rate, medium, device capacity, traffic load and protocol behaviour all affect performance. Throughput is the useful data rate actually achieved; latency is delay.
In a switched star, independent pairs can often communicate concurrently when the switch and links have capacity. However, many computers sending to one server can still compete for the server’s link. Dedicated access links do not create unlimited shared capacity.
For example, twenty computers may each have a 1 Gbit/s link, but if their traffic all crosses one 1 Gbit/s uplink, their combined useful throughput across that uplink cannot exceed its capacity. Extra links help only if the design can use them for the traffic in question.
A mesh can provide shorter or alternative paths, but extra hops, shared radio airtime, congestion or slower links can reduce performance. Do not write “mesh is always fastest” or “star is always slowest”. State the relevant conditions.
11 | DATA COLLISIONS: THE SHARING RULES MATTER
A data collision occurs when overlapping transmissions interfere on a shared medium in systems where that can happen. A collision domain is the region in which such competing transmissions can collide. This is not the same thing as general congestion.
With a legacy Ethernet hub, connected devices share the medium and half-duplex transmission can cause collisions. A physical star using a hub therefore behaves differently from a switched star, even when the cable drawing looks similar.
Correctly operating full-duplex switched Ethernet point-to-point links do not use collision-based access and do not experience normal Ethernet collisions. They can still suffer congestion, queueing and packet drops when too much traffic targets limited capacity.
Wireless devices contend for airtime, and overlapping transmissions can still cause loss despite collision-avoidance mechanisms. A wireless mesh is not collision-free simply because it has many possible routes. Compare the actual technology and access method, not just the topology name.
12 | SCALABILITY AND TOTAL COST
Scalability is the ability to support growth while maintaining acceptable service. A switched star is often easy to extend within its port and capacity limits. A hybrid can add local stars or strengthen selected core links. A full mesh needs many new connections for every new node.
Cost includes equipment, ports and cable, but also installation labour, building access, maintenance, power, monitoring and staff expertise. A low purchase price can be poor value if faults repeatedly interrupt important work.
A partial mesh often provides a practical compromise: add alternative routes for essential infrastructure, then keep ordinary endpoint access simpler. Choosing which links to add requires knowing traffic patterns and the impact of failure.
Do not assume that the most complex design is the best. A design that local staff cannot configure or maintain reliably can create its own outages.
13 | CHOOSE FOR THE SETTING
| Setting and requirement | Plausible design | Reason and remaining limitation |
|---|---|---|
| Home / small office: manageable budget and dependable local access | Switched star with suitable central equipment | Simple endpoint wiring and fault isolation; central unit and individual access links remain dependencies |
| Corporation: many departments and heavy shared-service traffic | Local stars with a resilient interconnected core | Manageable growth and selected alternative paths; uplinks and server connections still need sizing |
| Government service: disruption has a high impact | Hybrid with carefully planned partial mesh and diverse critical paths | Improves resilience to specified faults; configuration, power and recovery must be tested |
| College campus: several buildings and expanding demand | Star access networks with meshed or partially meshed backbone | Local simplicity plus resilient inter-building routes; building access switches may remain single points |
These are illustrative choices, not mandatory designs for every organisation of that type. A small medical practice may need stricter availability than a much larger non-critical teaching lab. Start from the requirement, then justify the topology.
14 | COMPARE THE FIVE REQUIRED FACTORS
| Factor | Star, using a modern switch | Mesh | Hybrid, such as star access plus mesh core |
|---|---|---|---|
| Reliability | Individual access faults usually isolated; central switch is critical | Alternative paths can survive selected faults; configuration and remaining connectivity matter | Can protect critical routes while leaving some local dependencies |
| Transmission speed | Concurrent links possible; central/uplink/server limits remain | Depends on paths, rates, routing and load; not automatically fastest | Can size access and core separately; bottlenecks still possible |
| Scalability | Add ports/links within capacity; larger expansion needs planning | Full-mesh connection growth is steep; partial mesh is more selective | Extend sections while strengthening core as needed |
| Data collisions | Full-duplex switched Ethernet avoids normal collisions; hubs differ | Depends on medium and duplex/access method | Depends on the technology in each component |
| Cost | Usually manageable for a small LAN; centre and cabling still cost | Full mesh has many connections and high management cost | Can target spending, but design and support complexity increase |
Use this table to check your understanding after reading the explanations. In an answer, connect a characteristic to the situation: “The office can add another computer using a spare switch port, so a switched star supports modest growth without rewiring every existing computer.”
15 | APPLY IT: MAKE A DESIGN DECISION
Read each situation and identify the most defensible statement. Some situations deliberately ask about limitations rather than naming a topology. The aim is to connect the arrangement to a requirement, not memorise a slogan.
16 | DESCRIBE THE FUNCTION AND ITS PRACTICAL VALUE
The command word describe asks for characteristics and main features. Name the arrangement, describe how its connections work, and show why those features suit the stated situation. Include a relevant limitation rather than claiming perfection.
For example: “A switched star connects the office devices through a central switch. A broken cable normally isolates only its attached device, so colleagues can continue working. Adding a device is straightforward when a spare port and sufficient capacity are available. However, failure of the central switch interrupts communication through that star.”
For a larger campus: “A hybrid can combine local stars with a partially meshed backbone. Local stars simplify endpoint connections, while alternative core paths can preserve connectivity after a selected link failure. This requires suitable forwarding configuration and costs more to install and maintain.”
When explaining consequences, use because, therefore or so that. Keep your claims conditional where the technology matters, particularly for speed, collisions and failover.