A network hub is a legacy layer-1 device that repeats an incoming data signal out of every other port without reading any address, and it is now obsolete. A hub joins several Ethernet devices into one shared segment and floods every frame to all connected ports at once. Because it reads no MAC address and makes no forwarding decision, a hub places every device in one collision domain and runs half-duplex, which wastes bandwidth and exposes traffic. The network switch replaced it on modern networks, and a new device sold as a hub is rare. For any network you build today, use a switch instead.
What Is a Network Hub?
A network hub is a physical-layer device that receives a data signal on one port and repeats that signal out of every other port. It connects several Ethernet cables into a single shared segment and treats all connected devices as one network.
A hub reads no MAC address and no IP address, so it cannot direct traffic toward a specific device. The IEEE 802.3 standard classifies a hub as a multiport repeater, because it regenerates the incoming signal and sends a copy to each remaining port. A hub and a repeater are functionally the same device; a repeater extends one cable, while a hub repeats to many ports at once.
A network hub differs from a switch and a router because it makes no forwarding decision. A switch reads the destination MAC address and sends a frame to one port, while a hub floods the frame to all ports. The role a hub plays in joining devices is the same role a network switch performs with targeted forwarding, which is the reason the switch replaced the hub.
How Does a Network Hub Work?
A network hub works by regenerating an incoming electrical signal and copying that signal to every other port. It operates only at the physical layer, so it handles raw bits rather than addressed frames.
The steps below trace a single transmission through a hub and show why the shared design produces collisions.
- A device sends bits. One connected device puts an electrical signal onto its cable, with no addressing the hub can read.
- The hub regenerates the signal. It restores the signal timing and amplitude, acting as a repeater rather than decoding a frame.
- It copies the signal to every other port. Each remaining device receives the transmission whether or not it was the intended recipient.
- A second sender causes a collision. If another device transmits at the same time, the two signals overlap on the shared segment and corrupt both frames.
- The hub propagates a jam signal. Under CSMA/CD it signals the collision to every port, so all devices back off and retransmit after a random delay.
Three rules summarize this behavior, and each one feeds directly into why the hub became obsolete:
- Broadcasts every signal. A hub copies each incoming bit stream to all other ports, so every connected device receives every transmission regardless of the intended recipient.
- Operates in half-duplex. A hub allows only one device to transmit at a time, so a device cannot send and receive data in the same instant.
- Forms one collision domain. A hub places all ports in a single collision domain, so two devices that transmit at once produce a collision that corrupts both frames.
A network hub relies on Carrier Sense Multiple Access with Collision Detection (CSMA/CD), defined in IEEE 802.3, to manage the shared medium. CSMA/CD requires each device to listen before transmitting and to back off after a collision. The shared collision domain limits a 10 Mbps or 100 Mbps hub to a fraction of its rated throughput once several devices transmit at the same time.
What Are Passive and Active Hubs?
A passive hub joins cables without amplifying the signal, while an active hub regenerates and boosts the signal before forwarding it. The distinction rests on whether the hub uses power to restore the signal.
- Passive hub. A passive hub connects ports through a wiring junction with no power and no signal regeneration, so it does not extend the cable distance limit.
- Active hub. An active hub draws power to regenerate the incoming signal, so it restores signal strength and extends the segment length, acting as a repeater.
- Intelligent hub. An intelligent hub adds management features such as port monitoring on top of active regeneration, so it reports basic status to an administrator.
A passive hub appears in older or low-cost installations, while an active hub serves where the cable run approaches the 100-meter limit defined for twisted-pair Ethernet in IEEE 802.3. Both hub types share the same broadcast and half-duplex behavior at the physical layer.
What Is the Difference Between a Hub and a Switch?
A hub broadcasts data to every port, while a switch forwards data to one port using the destination MAC address. A hub works at layer 1 and a switch works at layer 2 of the OSI model, which gives the two devices different performance.
The comparison table below lists the differences between a network hub and a network switch across the attributes that affect performance and design.
| Attribute | Network Hub | Network Switch |
|---|---|---|
| OSI layer | Layer 1 (physical) | Layer 2 (data link) |
| Forwarding method | Broadcasts to all ports | Forwards by MAC address to one port |
| Address table | None | MAC address table |
| Collision domains | One shared domain | One per port |
| Duplex mode | Half-duplex | Full-duplex |
| Bandwidth | Shared across all ports | Dedicated per port |
| Current use | Obsolete | Standard in modern networks |
A network switch builds a MAC address table by recording the source address of each frame, then uses that table to forward later frames to a single port. A hub keeps no table and floods every frame, so it wastes bandwidth and exposes traffic to all devices. The targeted forwarding of a switch is the reason a switch separates each port into its own collision domain. The wider device comparison is covered in switch vs router.
Why Are Network Hubs Obsolete?
Network hubs are obsolete because a switch forwards traffic faster, more securely, and without the collisions a hub creates. The shared design of a hub limits throughput and exposes data, both of which a switch resolves.
Three measurable problems pushed network hubs out of use. Each problem stems from the broadcast and half-duplex design of the hub.
- Wastes bandwidth. A hub shares one collision domain, so total throughput falls as devices are added and collisions rise under load.
- Exposes traffic. A hub sends every frame to every port, so any connected device can capture data meant for another, which a switch prevents.
- Forces half-duplex. A hub allows one transmission at a time, so it cannot reach the full-duplex speed a switch delivers on every port.
Manufacturing of standalone Ethernet hubs largely ended once the switch became cheaper to build. Modern Ethernet products listed under IEEE 802.3 use switching silicon, so a new device sold as a hub is rare. The shift mirrors the wider move described in how networks work, where addressed forwarding replaced shared broadcast.
Where Are Network Hubs Still Used?
Network hubs still appear in network analysis, legacy equipment, and teaching environments where shared broadcast is useful or where old hardware remains. The few remaining uses depend on the broadcast behavior that made the hub obsolete elsewhere.
- Packet capture. A hub copies all traffic to every port, so a technician can connect an analyzer to one port and capture every frame on the segment without a port mirror.
- Legacy systems. A hub may remain in an old industrial or lab setup where replacing the device would disrupt connected equipment.
- Teaching. A hub demonstrates collisions and the shared medium concept in a controlled lesson on CSMA/CD and Ethernet fundamentals.
Each remaining use relies on the hub broadcasting every frame. A managed switch with a port mirror feature, or a dedicated network TAP, now covers most capture needs, so even these uses decline as old hubs fail.
What Speeds and Standards Did Network Hubs Support?
Network hubs supported the 10 Mbps and 100 Mbps Ethernet standards, and no hub was made for Gigabit Ethernet. The shared, half-duplex design of a hub set a ceiling that the Gigabit standard could not meet, which marked the end of hub development.
The list below names the Ethernet standards that hubs supported and the limit that stopped hub development at higher speeds.
- 10BASE-T hubs ran at 10 Mbps. The earliest twisted-pair hubs followed the 10BASE-T standard in IEEE 802.3i, sharing 10 Mbps across every connected port.
- 100BASE-TX hubs ran at 100 Mbps. Fast Ethernet hubs followed IEEE 802.3u and shared 100 Mbps across the segment, still in half-duplex.
- Dual-speed hubs bridged 10 and 100 Mbps. A dual-speed hub held separate 10 Mbps and 100 Mbps segments joined by an internal bridge so mixed devices could connect.
- Gigabit Ethernet has no hub. The 1000BASE-T standard in IEEE 802.3ab requires full-duplex operation, which a shared half-duplex hub cannot provide.
The half-duplex requirement of CSMA/CD blocked any hub from reaching Gigabit speed, because Gigabit Ethernet assumes a full-duplex switched link. The throughput shared by a 100 Mbps hub fell well below its rating once several devices transmitted, which separates a hub from the dedicated per-port speed of a switch on the same cabling described in what Ethernet is.
Last Thoughts on Network Hubs
A network hub is a physical-layer device that repeats every incoming signal to all other ports, placing connected devices in one shared collision domain that runs half-duplex. It reads no address and makes no forwarding decision, which separates it from the addressed forwarding of a switch. That single shared design is the reason a hub wastes bandwidth, exposes traffic, and could never scale past 100 Mbps.
The targeted, full-duplex behavior of a network switch replaced the hub on modern networks, so a hub today is a teaching aid rather than a device you deploy. To see where these devices sit in the wider picture, the overview of what a computer network is connects the hub, the switch, and the router into one system.
Key Takeaways:
- A network hub is a legacy layer-1 repeater that copies a signal to every other port without reading any address.
- It broadcasts every frame to all ports, unlike a switch that forwards by MAC address to one port.
- All ports share one collision domain and run half-duplex, which limits throughput and causes collisions under load.
- A hub sends all traffic to all ports, so it is insecure; any connected device can capture another device’s data.
- The switch replaced the hub because it is faster, full-duplex, private, and about the same price.
- No Gigabit hub exists, and hubs survive only in packet capture, legacy gear, and teaching.
Frequently Asked Questions (FAQs)
What is a network hub in simple terms?
A network hub is a layer-1 device that copies an incoming signal out of every other port. It joins several Ethernet devices into one shared segment and broadcasts all data to every connected device, because it reads no address and makes no forwarding decision.
Is a network hub the same as a repeater?
Functionally yes. IEEE 802.3 classifies a hub as a multiport repeater, and the two terms are used interchangeably. A two-port repeater extends one cable, while a hub repeats the regenerated signal to many ports at once.
Why is a hub a security risk?
A hub sends every frame to every port, so any device plugged into it can capture traffic meant for another device. A switch forwards each frame only to the destination port, which removes that easy eavesdropping path.
Why was no Gigabit Ethernet hub ever made?
Gigabit Ethernet (1000BASE-T, IEEE 802.3ab) assumes a full-duplex switched link and drops the shared CSMA/CD model. A hub is half-duplex and shared by design, so it cannot meet the Gigabit standard, which ended hub development at 100 Mbps.
Should I ever buy a hub today?
No. A switch costs about the same, runs full-duplex, gives each port dedicated bandwidth, and keeps traffic private. Buy a switch for any network. A hub is worth keeping only as a lab or teaching tool.
What OSI layer does a hub operate at?
A network hub operates at layer 1, the physical layer of the OSI model. It handles raw electrical signals and reads no MAC or IP address, so it makes no forwarding decision and treats all ports as one segment.
