Network design and topologies define how devices and systems are physically and logically arranged, connected, and scaled within an enterprise or data center environment. Topology selection directly impacts performance, resilience, and growth potential—from simple bus configurations to multi-tier architectures handling thousands of devices. Modern network design balances redundancy, latency, and cost while accounting for traffic patterns (east-west vs. north-south), failure domains, and operational complexity. Understanding architectural patterns—three-tier campus, spine-leaf, collapsed core—and redundancy mechanisms like HSRP, dual-homing, and ECMP enables engineers to build networks that scale predictably, fail gracefully, and recover fast. Whether planning capacity for a growing WAN, segmenting security zones, or optimizing data center fabrics, the right topology and design principles are foundational to reliable, high-performance infrastructure.
What This Cheat Sheet Covers
This topic spans 12 focused tables and 91 indexed concepts. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.
Table 1: Fundamental Network Topologies
The basic shapes every network is built from — how nodes physically connect and where the traffic flows. Each pattern trades simplicity against resilience: a star is easy to wire but dies with its hub, a full mesh survives anything but explodes in cabling cost, and most real networks end up as hybrids that borrow the best of several. Knowing the failure mode of each shape is half of network design.
| Topology | Example | Description |
|---|---|---|
Central switch connects all devices in office LAN | • All nodes connect to a central hub or switch • hub failure brings down entire segment but individual device failures are isolated | |
Token Ring LAN or FDDI metropolitan network | • Data travels in one direction through a closed loop • single link failure breaks the ring unless dual-ring redundancy is implemented | |
Single coaxial cable backbone in early Ethernet | • All devices share one communication line • collision domain spans all nodes and cable break disconnects entire segment | |
Data center spine-leaf with all-to-all links | • Every device connects to every other device • provides maximum redundancy but scales poorly—requires n(n-1)/2 links for full mesh | |
WAN with strategic inter-site links | • Only critical nodes have multiple paths • balances redundancy with cost by connecting high-traffic pairs rather than all nodes |