A2.2.1 · Network architecture

Network Topologies

A home WiFi network with five devices and a hospital's life-critical monitoring network would make very different choices about how their devices actually connect to each other, and both choices would be right for their own situation.

Core for SL & HL Typically 3–5 marks · AO2 Command term: describe

Explainer

A network topology is the layout of how the devices (nodes) in a network are physically connected to each other.

Star topology

Every device connects only to one central device, usually a switch, which relays everything between them. This is the shape of most home and small-office networks.

Mesh topology

Every device connects directly to every other device. There is no single central point at all, which is exactly the point.

Hybrid topology

Two or more different topologies combined into one network, each part contributing the strength it's best at. A common shape: several switches meshed together to form a resilient backbone, each one then running its own star of ordinary devices.

Explainer

Comparing all five factors

The statement names five specific factors a comparison must cover. A "describe" answer that only mentions cost, or only mentions reliability, is incomplete.

Factor	Star	Mesh	Hybrid
Reliability	Whole network fails only if the central switch fails; a single device failing affects only that device	Very high, no single point of failure at all, traffic reroutes around any broken link	High, a failure usually only affects the local section it happened in
Transmission speed	Good, no device competes with another for its own private line, but whatever the central switch can handle sets the overall ceiling	Can be very fast, but depends on how many hops data has to take between distant nodes	Depends entirely on the specific mix chosen, poor design anywhere can create a bottleneck
Scalability	Limited by the switch's fixed number of ports	Expensive to extend, new cabling and connections are needed to every existing node	Generally the most flexible, since new devices can join whichever section suits them best
Data collisions	Low, since each device's traffic travels down its own private line to the switch, not shared with anyone else's	Low, direct connections mean less competition for the same path	Depends on the specific mix, but generally kept low by the same properties as its component topologies
Cost	Low for a handful of devices, since only short runs to one central point are needed	High, every device needs its own cable run to every other device it connects to	Varies widely, often higher overall due to combining and managing more than one design

Worked example

Matching a topology to a real setting

The command term is "describe... practical applications," and a topology's real value only shows up once it's actually applied to a setting with genuine constraints.

Choosing the right topology for the job Step 1 of 3

Setting 1: a family's home network. Eight devices total, a modest budget, and a single WiFi router already sitting in the lounge.

Star. Eight devices is comfortably within a typical router's port and connection capacity, and the cost of a star setup at this scale is genuinely low, just one central device and a short cable or WiFi link to each device. Reliability is fine too: the only real risk is the router itself failing, which is an acceptable trade-off for a home.

Setting 2: a hospital's critical patient-monitoring network. Losing connectivity for even a few seconds during a procedure is unacceptable, no matter the extra cost.

Mesh. This is exactly the scenario mesh exists for: no single point of failure means no single equipment fault can take the whole monitoring network down. The much higher cabling cost is a reasonable price for a setting where reliability genuinely outweighs everything else.

Setting 3: a college campus spanning several buildings. Thousands of devices overall, but manageable numbers within any one building, and the whole campus needs to keep functioning even if one building's own network hiccups.

Hybrid. A meshed backbone linking the buildings together gives the whole campus resilience against any one building's connection failing, while each individual building simply runs its own ordinary, cost-effective star for its own devices. Neither a pure star (not resilient enough at this scale) nor a pure mesh (wildly expensive to cable every single device to every other) fits as well as combining both.

Check your understanding

A national air-traffic-control network absolutely cannot afford total connectivity loss if any one piece of equipment fails, and cost is a secondary concern compared to that requirement. Which topology best fits, and why?

Common mistake

A recurring confusion this statement's own emphasis on visualising node failure is written to guard against

RecurringA common weak answer claims that in a star topology, ANY device failing brings down the whole network. This is only true of the central switch. If one of the ordinary connected devices fails, every other device is completely unaffected, only that one device is cut off.

Overstates the risk: "A star topology is unreliable because if any device fails, the whole network goes down." True only for the central device; treats a completely different, much less serious failure as if it were the same thing.

✓

Distinguishes the two cases: "In a star topology, if an ordinary device fails, only that device is affected. If the central switch fails, the entire network goes down, since every device's connection runs through it." Correctly separates the two very different failure scenarios, which is what the statement's own emphasis on visualising node failure is actually testing.

Linking question

How this connects to the rest of the syllabus

The IB syllabus is deliberately woven together, not a stack of isolated topics. Here's how network topologies connect specifically to A4 (machine learning), worth thinking through, not just memorising: it's the kind of synthesis a Paper 1/2 question can genuinely ask for.

How do network topologies influence machine learning algorithms? (Training a machine learning model across many machines at once, distributed training, depends heavily on how those machines are connected: a mesh-like arrangement lets machines exchange updates directly and often trains faster, while a star-like arrangement funnels everything through one central coordinator, which can become the exact same kind of bottleneck a star topology has for ordinary network traffic.)

Revision questions

Past-paper style. Write an answer, then reveal the marking points and mark yourself.

Question 1 · 2 marks

[2]Describe what happens to a star network if a single ordinary device fails, compared to if the central switch fails.

Question 2 · 3 marks

[3]Explain why a mesh topology is generally more expensive to scale than a star topology.

Question 3 · 4 marks

[4]A growing company is expanding from one office to three separate office buildings on the same street. Recommend a suitable network topology and justify your answer.

Flashcards

Active recall for this statement’s key terms.