Network Hub vs Switch: What Is the Difference?

A network hub is a simple Ethernet device that connects several computers into one network segment by repeating every signal it receives out of all its other ports. It works at layer 1 of the OSI model, the physical layer, and has no idea which device a frame is addressed to. The IEEE 802.3 standard calls it a repeater, and an Ethernet hub is the same device under its common name.

Because every frame goes everywhere, all devices on a hub share one collision domain. Only one can transmit at a time, so links run half-duplex and rely on CSMA/CD (carrier sense multiple access with collision detection) to back off and retry when two devices transmit at once. A 100 Mb/s hub with ten busy computers gives each of them a share of 100 Mb/s, not 100 Mb/s each.

Hubs were common in 10 and 100 Mb/s networks of the 1990s. Switch prices then fell below hub prices, IEEE stopped developing repeaters, and hubs were never made in any quantity for gigabit Ethernet.

A switch works at layer 2, the data link layer. It reads the source MAC address of each frame arriving on a port and records it in a MAC address table. When a frame arrives for a known address, the switch sends it only out of the port where that device lives. Unknown and broadcast frames are flooded, but that is the exception rather than the rule.

Hub networks also had strict size limits. On 10 Mb/s Ethernet the 5-4-3 rule allowed at most five segments joined by four repeaters, with only three segments carrying computers, so that collisions could still be detected across the whole network. Fast Ethernet was tighter still, typically allowing only one or two hubs between any two devices. A switch removes those limits because it buffers and re-sends each frame, so switches can be chained, stacked and meshed. The products sold as “switching hubs” in the late 1990s were simply early switches.

Each switch port is its own collision domain and runs full-duplex, so every device can send and receive at full line rate at the same time. Our explainer on how an Ethernet switch forwards traffic covers MAC learning and switching methods in more depth.

The table compares a hub with the two kinds of switch you are likely to buy today.

An unmanaged switch is the direct replacement for a hub. It needs no setup and plugs in the same way, but forwards traffic far more efficiently.

The three devices are often confused because they look alike: a box with a row of RJ45 ports.

Old hubs still turn up in machine panels and legacy control cabinets, usually at 10 Mb/s. Replacing one with a switch removes collisions, gives each controller its own bandwidth and lets ports run full-duplex. Check that older devices auto-negotiate correctly first, as a few fixed half-duplex devices need their switch port forced to match.

For a cabinet, unmanaged industrial switches are the like-for-like swap, with DIN-rail mounting, wide-range DC input and wide operating temperatures. Where the network needs VLANs, redundancy or monitoring, a managed switch is the better fit, and our guide to choosing between managed and unmanaged industrial switches sets out the trade-offs. Engineers who used a hub to capture traffic for diagnostics can use port mirroring on a managed switch instead.

We supply industrial Ethernet switches with the port count, PoE budget, mounting and power input each installation needs, and configure managed switches as standard before they ship. Send us your device list and we will specify the switch to match.