hub-and-spoke

cloudnetworking

The topology where everything connects through a central hub instead of to each other - fewer links, central control, one very important middle.

The design behind transit gateways, SD-WAN backbones, and most airline maps.

Hub and spoke is a topology where sites connect to a central location rather than to each other, and every conversation between two spokes passes through the hub. It was the default enterprise design for decades because the hub was where the applications, the internet connection and the security controls lived.

Its virtue is control and its cost is the path. Traffic from one branch to another travels to the centre and back, which adds latency proportional to how far the centre is from both, and the hub becomes both a bottleneck and a single point of failure that requires expensive redundancy. When the applications were genuinely in the data centre, the extra distance was buying something real.

That justification weakened when the applications moved. If the destination is a cloud service, backhauling a branch's internet traffic to a distant hub in order to inspect it is paying latency for a security control that could be applied closer, which is the entire argument behind SD-WAN and the SASE model. The topology is not obsolete, since regulated traffic and internal systems still justify it, but it is now a choice rather than the assumption.

← All glossary entries