Skip to content
FALCON CORETECHNOLOGIES

// Lesson 5 of 49 · Network fundamentals

Network topologies and architectures

Two-tier and three-tier campus designs, spine-leaf data centres, WAN and SOHO networks, and on-premises vs cloud.
2 min read#design#campus#spine-leaf#cloud

Real networks follow a small number of proven designs. Recognising them tells you where a device sits and what it is responsible for.

Three-tier campus

Layer Role
Access Where endpoints connect. VLANs, PoE, port security, edge protection.
Distribution Aggregates access switches. Inter-VLAN routing, policy, the boundary between Layer 2 and Layer 3.
Core Fast, simple, highly available transport between distribution blocks and to the data centre / WAN.

Two-tier (collapsed core)

Smaller sites combine core and distribution into one pair of switches. It is cheaper and simpler and works well until the site grows large.

Spine-leaf (data centre)

  • Every leaf switch connects to every spine switch; leaves never connect to each other, nor spines to each other.
  • Any server is the same number of hops from any other server: leaf → spine → leaf.
  • Capacity grows by adding spines (more bandwidth) or leaves (more ports).

This design suits the heavy east-west traffic (server to server) of modern data centres.

WAN

A WAN connects sites over long distances, through a service provider or the internet: MPLS, dedicated links, broadband, 4G/5G and SD-WAN (which manages several of these centrally and chooses paths per application).

SOHO

A small office / home office network often uses one device that is a router, switch, firewall, access point and DHCP server at the same time.

On-premises vs cloud

  • On-premises: you own and operate the hardware.
  • Cloud: a provider runs the infrastructure; you consume it as a service.
Model The provider manages You manage
IaaS Hardware, virtualisation OS, applications, data
PaaS Up to the runtime Applications, data
SaaS Everything Your data and users

Key points

  • Campus: access → distribution → core; small sites collapse the core.
  • Spine-leaf gives predictable latency for server-to-server traffic.
  • SD-WAN centrally manages multiple WAN links.
  • IaaS, PaaS and SaaS differ in how much the provider manages.

Sign in to save your lesson progress across devices. You can read every published lesson without an account.