Network Topologies

Part of Networking

The physical and logical arrangement of nodes and connections in a network.

Why This Matters

The topology of a network — how nodes are arranged and interconnected — determines its cost, performance, reliability, and scalability. A topology that works for five devices in one room may fail completely when scaled to fifty devices across a building. The wrong topology creates bottlenecks, single points of failure, and maintenance nightmares. The right topology, chosen to match actual requirements, enables a network to grow and evolve without requiring complete redesign.

Understanding topologies lets you read and create network diagrams, evaluate proposed network designs, identify why a network is failing or performing poorly, and design networks that will work reliably for years. In a reconstruction context, where resources for network infrastructure are limited and reliability is critical, choosing the right topology from the start avoids expensive rework.

Bus Topology

In a bus topology, all devices connect to a single shared cable. Data transmitted by any device propagates along the entire cable and is received by all other devices. Each device examines the destination address of each frame and processes only frames addressed to itself.

Bus topology was used in early Ethernet networks (10Base2 with coaxial cable). Its advantages are simplicity and low cost — only one cable runs through the entire installation, and each device connects with a simple T-connector and terminators at each cable end. Adding a device requires only inserting a T-connector at any point along the cable.

The disadvantages are significant. The entire network shares a single collision domain — every transmission competes with every other, and CSMA/CD manages access. Performance degrades rapidly as device count increases. A single cable break anywhere disrupts the entire network. The terminator resistors at each cable end are easy to forget and cause network failure when missing.

Bus topology is essentially obsolete for new installations but may be encountered in older buildings. If you inherit a 10Base2 coaxial network, be aware of its fragility: any connection problem anywhere on the cable disrupts everything. Converting to star topology is worth the investment.

Star Topology

In a star topology, each device connects with its own dedicated cable to a central hub or switch. No device connects directly to another device — all traffic flows through the central device.

Star topology is the dominant physical topology for modern Ethernet. Its advantages are numerous: a cable failure affects only the one device on that cable; adding a device requires only running a new cable to the central switch; troubleshooting is straightforward because each link is independent; the central switch provides visibility and control through its management interface.

The disadvantage is cost and complexity of the central device and additional cable runs. A hub-based star network costs more than a bus network of equivalent scale. The central device is also a single point of failure — if the central switch fails, the entire network goes down. Redundant switches connected with uplinks mitigate this risk.

For most installations, star topology with a switch is the obvious choice. The benefits in reliability and maintainability far outweigh the additional cost, especially at any scale larger than two or three devices.

Ring Topology

In a ring topology, each device connects to exactly two neighbors, forming a closed loop. Data travels around the ring in one direction (or both directions in a bidirectional ring). Each device receives frames from its upstream neighbor and retransmits them to its downstream neighbor.

Token Ring networks used this topology. A special control frame (the token) circulates around the ring; only the device holding the token may transmit. This eliminates collisions at the cost of introducing token management complexity and the potential for token loss (if the device holding the token fails).

SONET/SDH fiber optic rings use a self-healing ring architecture where each node connects to two separate ring paths. If one path fails, traffic automatically reroutes over the other path. This provides very high availability at the cost of doubling the fiber infrastructure.

Ring topology is uncommon in local networks today but is relevant in metropolitan and wide-area fiber networks where self-healing is valuable and the cost of redundant fiber can be justified by the reliability benefit.

Mesh Topology

In a full mesh topology, every node connects directly to every other node. A network with N nodes has N(N-1)/2 links. Five nodes require 10 links; ten nodes require 45 links. Full mesh provides maximum redundancy — any single link or single node can fail without disrupting communication between any remaining nodes — but the exponential cost makes it impractical for more than a few nodes.

Partial mesh topology connects each node to several (but not all) other nodes. This provides redundancy at lower cost by accepting that some failure scenarios will disrupt some paths. The design tradeoff is determining which nodes need redundant connections and which do not.

Mesh topology is used in backbone networks connecting major routers, in wireless mesh networks (where every device can relay traffic for other devices), and in data center fabrics connecting large numbers of servers. In each case, the multiple paths are exploited by routing protocols that automatically use alternate paths when primary paths fail.

Hierarchical (Tree) Topology

Real networks are typically hierarchical, combining multiple topologies at different scales. A campus network might use: a mesh of connections between core switches (for redundancy), a star of switches connecting floors or buildings to the core, and star segments connecting individual devices to floor switches.

This hierarchy allows the network to scale — the access layer (connecting individual devices) uses simple, inexpensive switches; the distribution layer (connecting access switches) uses mid-range switches with more capacity; the core (connecting distribution switches) uses high-performance switches with maximum throughput.

Spanning Tree Protocol (STP) is essential in hierarchical networks with redundant links. Without STP, loops in the topology create broadcast storms that saturate the network. STP automatically detects loops and blocks some paths to create a loop-free logical topology while keeping the redundant physical links available as standby paths.

Logical Topology vs. Physical Topology

Physical topology is the actual arrangement of cables and devices. Logical topology is how traffic flows, which may differ from the physical arrangement. A star-wired Ethernet hub has a physical star topology but a logical bus topology — all devices share a single collision domain regardless of the separate cable runs.

VLANs (Virtual LANs) further separate physical and logical topologies. Multiple VLANs can exist on a single physical switch, creating logically separate networks. Devices on different VLANs cannot communicate at Layer 2 even if they are physically connected to the same switch — they require a router to communicate. This allows network segmentation (separating departments, guest networks, or sensitive systems) without purchasing separate physical equipment.

Choosing a Topology

For a new network installation, the decision process is straightforward:

Start with star topology using switches. This is the correct choice for nearly every installation. Run a cable from each device location to a central wiring closet, terminate on a patch panel, and connect patch panels to switches.

Add redundancy proportional to the importance of reliability. For critical connections (servers, the connection between buildings), add redundant links. For user desktops, a single link is sufficient — the inconvenience of a failed desktop link is minor compared to the cost of redundant infrastructure.

Use hierarchy for scale. A single switch handles up to 48 devices. Larger networks use a hierarchy of switches: edge switches connecting individual devices, distribution switches connecting edge switches, and (for very large networks) core switches connecting distribution switches.

Document the topology. A network diagram showing every device, every link, and every connection point is essential for maintenance. Without documentation, diagnosing problems and making changes becomes guesswork. Update the diagram whenever the physical network changes.