Topology Comparison
The same network has two topologies at once: the physical one you can see with cables, and the logical one data actually follows. This lesson defines both and then compares the classic shapes — point-to-point, star, mesh, bus, and ring — by their advantages, disadvantages, and typical use.
Physical vs logical topology
Every network has two answers to the question “how is it laid out?” They are not the same, and confusing them is a common exam trap.
The two can disagree completely. A modern office is wired as a physical star: every PC runs a cable to a central switch. But the logical picture is different again — a switch forwards frames only where they need to go, and VLANs can carve the building into several logical networks that ignore the physical cabling. A building physically wired as a star can behave logically like a set of separate networks, or even like a ring in some legacy token-ring deployments.
Point-to-point, star and hub-and-spoke
Two shapes cover most of what you will build or read about.
Point-to-point
A point-to-point link joins exactly two devices and nothing else. There is one path, dedicated to those two endpoints, so bandwidth is guaranteed and there is no sharing or contention. Point-to-point links are the building block of WANs: a serial link between two branch routers, or a leased line from a site to a provider, is a point-to-point connection.
- Advantage: simplest possible topology; predictable, dedicated bandwidth.
- Disadvantage: it scales only in pairs — connecting many sites this way becomes a mesh and the link count explodes.
- Typical use: WAN connections between two sites; the fundamental link inside most other topologies.
Star and hub-and-spoke
In a star topology every device connects to one central device (a switch, hub, or access point). All traffic passes through the centre. Hub-and-spoke is the WAN cousin: a central hub site with spoke sites attached to it, where the centre is typically a router rather than a switch.
- Advantage: easy to add or remove a device — just cable it to the centre; centralising management and policy; a single faulty device or cable affects only its own segment.
- Disadvantage: the central device is a single point of failure and a potential bottleneck, since every frame must pass through it.
- Typical use: virtually every modern Ethernet LAN (switches create a physical star), wireless networks (the access point is the centre), and branch offices (hub-and-spoke WAN).
Full mesh and partial mesh
In a full mesh, every node connects directly to every other node. If there are n nodes, the number of links is n(n−1)/2 — because each of the n nodes links to the other n−1, and each link is counted twice.
- Advantage: the highest possible redundancy and performance — if any one link fails, another path exists; and every pair of nodes has a direct route.
- Disadvantage: the cost and port count grow quickly with each node added; it becomes unmanageable for large networks.
- Typical use: small, high-value cores — for example three or four critical routers, or the links between the tiers of a spine-leaf fabric.
Partial mesh
A partial mesh is a compromise: some nodes are fully interconnected, while others connect to only one or two neighbours. It delivers most of the redundancy of a full mesh at a fraction of the cost, because only the busiest or most critical links are doubled.
- Advantage: a good balance of redundancy, performance, and cost; no single link failure need isolate any site if the mesh is designed well.
- Disadvantage: more complex to plan and route than a hub-and-spoke; routing must be configured so traffic takes the intended path.
- Typical use: most real-world WANs and large campus cores, where a few key sites are meshed and the rest connect simply.
Bus and ring
Two shapes that explain a lot of exam questions — because they are legacy, and because their weaknesses are severe.
Bus
A bus topology uses a single shared backbone cable (the bus) to which every device taps in; both ends of the cable are terminated. Any signal put on the bus is heard by all devices.
- Advantage: very cheap and simple to wire for a small number of nodes; needs no central device.
- Disadvantage: a break in the backbone (or a missing terminator) brings the whole network down; all devices share one collision domain, so performance collapses as nodes are added; it is hard to troubleshoot.
- Typical use: legacy Ethernets (10BASE5 and 10BASE2). It is not used in modern switched LANs and appears mainly as a conceptual and historical question.
Ring
In a ring, each device connects to exactly two neighbours, forming a closed loop. Data travels around the ring from node to node; in a dual ring there are two paths so that a single break can be routed around.
- Advantage: predictable, orderly access to the medium; a dual ring tolerates a single link or device failure without losing connectivity.
- Disadvantage: in a single ring, a break interrupts the loop and takes the network down; adding or removing a node can disrupt traffic; higher latency as data passes through intermediate nodes.
- Typical use: legacy Token Ring and FDDI LANs, and some service-provider or metropolitan rings (for example SONET/SDH) where the dual-ring resilience is valuable.
The comparison table
This table summarises everything above. Learn it as a set of trade-offs, not as separate facts: the one theme running through the whole list is that redundancy costs money and complexity, while cheap and simple means fewer safety nets.
| Topology | Advantages | Disadvantages | Typical use |
|---|---|---|---|
| Point-to-point | Simplest; dedicated, predictable bandwidth; one path only. | Links only two nodes; many sites become a mesh. | WAN links between two sites; leased lines. |
| Star / hub-and-spoke | Easy to add or remove nodes; centralised management; a bad node affects only its own segment. | Central device is a single point of failure and a bottleneck. | Switched Ethernet LANs, wireless, branch WANs. |
| Full mesh | Maximum redundancy and performance; direct path between all pairs. | Costly; n(n−1)/2 links; hard to manage as it grows. | Small critical cores; spine-leaf tier links. |
| Partial mesh | Good redundancy at moderate cost; suited to real budgets. | More complex routing and planning than a star. | Most enterprise WANs; large campus cores. |
| Bus | Cheap and simple for a few nodes; no central device. | One break downs the whole network; shared collision domain; hard to troubleshoot. | Legacy Ethernets (10BASE5/10BASE2); historical only. |
| Ring | Orderly access; a dual ring survives one failure. | A single-ring break drops the network; latency grows through intermediate nodes. | Legacy Token Ring/FDDI; provider metro rings. |
Choose a topology
The last skill is matching a requirement to a shape. The table below is the revision sheet for that decision.
| Requirement | Best-fit topology | Why |
|---|---|---|
| Every PC connects to one central switch | Star | Adding or removing a PC touches one cable, and a faulty PC affects only its own segment. The centre remains a single point of failure, so production designs back it up. |
| 20 branches on a limited budget | Hub-and-spoke or partial mesh | A full mesh would need 190 links. A hub-and-spoke costs 19; a partial mesh adds a few direct links between the busiest sites to remove the hub as a single point of failure. |
| 3 critical routers | Full mesh | n(n−1)/2 = 3×2/2 = 3 links. Any one link can fail and traffic still flows, and each router has a direct path to both others. |
| An old coax backbone | Bus | A shared backbone with terminators at both ends. Cheap and simple, but one break takes down every device on it, so it survives only in legacy or very small installations. |
Review
Check what you have learned with these review questions.
Sources & references
- Cisco. CCNA (200-301) Exam Topics, Domain 1.0 Network Fundamentals. cisco.com.
- IEEE. IEEE 802.3 Ethernet Standard (legacy bus and star physical topologies). standards.ieee.org.
- IEEE. IEEE 802.5 Token Ring Standard (ring topology). standards.ieee.org.
- Tanenbaum, A. S. & Wetherall, D. J. Computer Networks, 5th ed., Chapter 1 (network topologies and classification).
Target exams: CCNA 200-301 v1.1 · CyberOps Associate 200-201. Last content verification: 2026-10-11.