What Is Mesh Topology?

Mesh topology is a network layout in which each device connects to several others, so there is more than one path between any two points. If one link or node fails, traffic takes another route. A mesh network can be wired, with switches and routers linked by cable or fibre, or wireless, with radios relaying traffic for each other.

Because a mesh has loops, it needs a protocol that chooses paths and stops traffic circling. Wired meshes use routing protocols such as OSPF at layer 3, or layer 2 protocols such as Shortest Path Bridging (IEEE 802.1aq) and the Multiple Spanning Tree Protocol in IEEE 802.1Q. Wireless meshes use their own path selection, for example the HWMP protocol in IEEE 802.11s.

Each node in a mesh relays traffic for others as well as sending its own. That is what lets a mesh route around damage, and it is also why a mesh is harder to plan and troubleshoot than a star.

A full mesh links every node directly to every other node. A partial mesh adds redundant links only between the nodes that need them. The difference shows most clearly in the link count, which grows quickly in a full mesh.

Because the link count of a full mesh grows with the square of the number of nodes, full meshes stay small. Most real networks are partial meshes: a well-linked core with simpler layouts hanging off it.

A mesh buys resilience with cabling, ports and configuration effort.

Wired meshes are built from managed switches and routers. In a data centre, a spine-leaf design links every leaf switch to every spine switch, a partial mesh that gives each server rack several equal paths. In a campus or plant backbone, core switches are cross-linked so the loss of one does not isolate a building.

Routing at layer 3 is the usual way to run a large wired mesh, because routing protocols use every link rather than blocking the redundant ones. Our guide to when you need a layer 3 switch explains that choice, and our comparison of layer 3 switches and routers covers the overlap with routers. Smaller layer 2 meshes can run on managed industrial switches using MSTP to block redundant links until they are needed.

Wireless meshes suit sites where cable is impractical. Several standards use them:

Few plants run a true full mesh. The usual pattern is a partial mesh at the core, where two or more backbone switches are cross-linked, with rings or stars of switches feeding the cabinets on the plant floor. That keeps the redundant paths where an outage would stop the most equipment, and keeps the edge simple to maintain.

When designing one, decide first which failures the network must survive: a single cable, a whole switch or a whole cabinet. Then add links only where they cover that failure, choose one protocol to manage them, and document which links are blocked or preferred. Wireless meshes on outdoor sites need a radio survey, because each hop shares airtime and the slowest link sets the pace for everything behind it.

We supply the managed and layer 3 switches for meshed backbones and configure the redundancy protocols, VLANs and routing before the system ships.