Question 5 of 7: Intranet Network Topologies

Question 5: Intranet Network Topologies (20 marks)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Connecting ten machines into a shared-resource intranet is fundamentally a choice of physical/logical topology — the pattern in which the links between the ten nodes are arranged. The four classical LAN topologies below are all viable candidates, and a real installation for ten hosts would most likely end up as a variant of the star (the dominant choice in modern Ethernet/switched LANs), but each is discussed on its merits since the question asks for potential configurations generally.

Bus

All ten machines tap onto a single shared cable, with terminators at each end. Advantages: minimal cabling (cheapest to install for a small, physically compact office), easy to extend by adding another tap. Drawbacks: the whole network is a single collision domain, so throughput degrades sharply as more of the ten machines transmit; a single cable break or a bad terminator takes the entire network down; and fault isolation is difficult (any tap could be the culprit).

Star (via a central hub/switch)

Each of the ten machines runs its own cable to a central hub or switch. Advantages: a broken cable or failed NIC affects only that one machine, not the whole company; a switch (versus a hub) gives each link its own collision domain, so ten machines can approach full aggregate throughput simultaneously; centralizing at the hub also makes it the natural place to add shared resources (a file/print server) and manage/monitor the whole intranet. Drawbacks: the central hub/switch is a single point of failure for the entire network, uses more cable overall than a bus, and adds the cost of the central device.

Ring

The ten machines are connected in a closed loop, each talking only to its two neighbours, with data (and, in token-ring designs, a token) passed around the ring. Advantages: access is orderly (no collisions when using token-passing), and performance degrades gracefully and predictably as load increases, unlike bus contention. Drawbacks: a single broken link or failed node can take down the whole ring unless a dual-ring or self-healing design is used (extra cost); adding or removing a machine disrupts the ring; and total delay grows with the number of machines the data must pass through.

(Partial) Mesh

Some or all pairs of the ten machines are connected directly to each other; a full mesh of ten nodes would need $\binom{10}{2}=45$ links. Advantages: maximum redundancy and fault tolerance (no single link or node failure isolates the network) and the highest possible aggregate throughput, since traffic between any pair need not compete for a shared medium. Drawbacks: cabling and port cost scale as $O(n^2)$, which is prohibitive for a full mesh of ten machines (45 links); in practice only a partial mesh (redundant links between the most critical machines only) is affordable at this scale.