Network Types and Topologies

Networks can be classified in two complementary ways: by their geographic scale (how far the network stretches) and by their topology (how the devices are physically or logically arranged). Understanding both classifications helps engineers choose the right hardware, cabling, and design for a given situation, whether that's a smartwatch talking to a phone or a multinational company connecting offices across continents.

By scale, a PAN (Personal Area Network) covers a very small range, typically a few meters, connecting personal devices like a phone, smartwatch, and wireless earbuds via Bluetooth. A LAN (Local Area Network) covers a single building or campus, such as an office or home network, usually built with Ethernet and Wi-Fi and owned entirely by one organization. A MAN (Metropolitan Area Network) spans a city or large campus, often used by municipalities or ISPs to connect multiple LANs across a metro area. A WAN (Wide Area Network) spans countries or continents — the internet itself is the largest WAN — and typically relies on leased lines, fiber backbones, or satellite links owned by multiple providers. As scale increases from PAN to WAN, latency generally increases, and the network typically spans more administrative boundaries and ownership domains. By topology, a bus topology connects all devices to a single shared cable; it is cheap and simple to cable but a single break in the backbone cable disables the whole segment, and performance degrades as more devices contend for the shared medium. A star topology connects every device to a central hub or switch; it is easy to install and troubleshoot, and a single cable failure only affects one device, but the central device is a single point of failure and requires more cabling than a bus. A ring topology connects devices in a closed loop, with data passing from one device to the next; it can offer predictable performance, but a single break can disable the entire ring unless a dual-ring or self-healing design is used. A mesh topology connects devices with many redundant point-to-point links; full mesh offers the highest fault tolerance because there are multiple paths between any two nodes, but it is the most expensive and complex to cable, since the number of links grows quadratically with the number of nodes. A hybrid topology combines two or more of these designs — for example, a star-of-stars or a star combined with a mesh backbone — to balance cost, fault tolerance, and manageability for large real-world networks.

The mesh_links calculation illustrates the fundamental cost tradeoff of mesh topology: fault tolerance is excellent because there are many alternate paths, but cabling cost explodes quadratically as the network grows, which is why full mesh is rarely used beyond small core networks (such as connecting a handful of routers between data centers). In contrast, star topology scales linearly with the number of devices (one cable per device to the central switch), which is why it dominates modern LAN design — the tradeoff is that everything depends on the central switch staying operational. Bus and ring topologies are mostly of historical or specialized interest today (early Ethernet used bus wiring; some industrial and metro networks still use ring designs for their predictable failover behavior), while hybrid topologies — such as a campus network built from many star-wired wiring closets linked by a resilient mesh or ring backbone — are what most real, large-scale networks actually use in practice.