Course contents

Unit 1: Networking BasicsLesson 1.5 (5 of 5 in this unit)5 of 104 in the Network Fundamentals course

WAN Topology Types

A LAN topology describes cables in one building. A WAN topology describes the links between sites. The choices look familiar, but the cost per link is much higher, so the trade-offs bite harder.

Beginner · 9 min read · Before this: Types of networks, Network topologies

Covers Point-to-point, Hub-and-spoke, Partial and full mesh, Link counts and Choosing a WAN shape.

Technical definition

A WAN topology is the pattern of links that joins separate sites, such as offices, data centres and branches, across a wide area. It decides cost, resilience and the path traffic takes between sites.

WAN topologies: how sites are joined
WAN topologies: how sites are joinedFour ways to join a head office (HQ) and branch sites over a WAN. Point-to-point: one dedicated link between two sites. Hub-and-spoke: every branch links to the HQ hub, so branch traffic to other branches passes through HQ. Partial mesh: some extra direct links between branches. Full mesh: every site links to every other, with the best resilience and the highest cost.Point-to-pointone link, two sitessimple; the link is the riskHQB1Hub-and-spokeall traffic via the hubcheap; hub is the riskHQB1B2B3B4Partial meshextra links where neededbalances cost and resilienceHQB1B2B3B4Full meshevery site to every sitebest paths; most links to buyHQB1B2B3B4

On small screens, swipe sideways or focus the diagram and use the arrow keys.

More links mean more resilience and more cost. A full mesh of n sites needs n(n-1)/2 links: 4 sites need 6, 10 sites need 45.

Point-to-point

One dedicated link joins two sites, such as a leased line between a head office and a branch. It is simple and predictable. The link is the single point of failure, and it does not scale: each new site needs its own link.

Hub-and-spoke

Every branch has one link to a central site, the hub. This needs only as many links as there are branches, which keeps it cheap, and it puts policy and security in one place. Traffic between two branches passes through the hub, which adds delay, and if the hub fails, every branch loses its path to the others and to central services.

Partial mesh and full mesh

A partial mesh adds direct links between selected sites, usually where traffic is heavy or resilience matters. A full mesh connects every site to every other, which gives the shortest paths and the most resilience at the highest cost.

Links needed for a full mesh
Links needed for a full meshSites; Links in a full mesh; Links in hub-and-spoke. 4: 6 3 10: 45 9 20: 190 19 50: 1,225 49SitesLinks in a full meshLinks in hub-and-spoke463104592019019501,22549

On small screens, swipe sideways or focus the diagram and use the arrow keys.

A full mesh of n sites needs n(n-1)/2 links; hub-and-spoke needs n-1.

Choosing a WAN shape

Match the topology to where traffic really flows. If branches mostly talk to head office and data centres, hub-and-spoke fits. If branches talk to each other a lot, add direct links where that traffic is. Reserve full mesh for a handful of critical sites. Whatever shape you choose, give important sites a second path, because the WAN link is the part you do not own. The same shapes appear again with VPNs and SD-WAN, where tunnels replace leased lines.

Test yourself

Predict · scenario 1

A company has 10 branches, each linked only to head office. Which topology is it?

Predict · scenario 2

How many links does a full mesh of 6 sites need?

Predict · scenario 3

What is the main weakness of hub-and-spoke?

FAQ

Is hub-and-spoke still used today?
Yes. It is simple and cheap, and many VPN and SD-WAN designs build on it. Direct branch-to-branch tunnels are often added where traffic between branches is heavy.
How many links does a full mesh need?
n(n-1)/2 for n sites: 4 sites need 6 links, 10 sites need 45, and 50 sites need 1,225. That growth is why full meshes stay small.
What is a dual-homed site?
A site with two connections, for example to two providers or two hubs, so one failure does not cut it off. It adds resilience without a full mesh.