VMware ESXi and vSphere Cluster Management

What Is a Network Hub?

Learn how Ethernet hubs work at OSI Layer 1, how they repeat signals, why they create shared collision domains, and how hubs differ from switches.

A network hub is a legacy Ethernet device that connects multiple hosts to one shared local area network (LAN) segment. A LAN is a network that connects devices within a limited area, such as a home, office, classroom, or building.

A hub is a multiport repeater. It receives an electrical or optical signal on one port, regenerates that signal, and sends it through all of its other active ports. It does not decide which connected device should receive the traffic.

Hubs were once used to connect computers in small Ethernet networks. Modern LANs generally use switches instead because switches can forward traffic selectively, support full-duplex communication, and provide better performance and traffic separation.

How a Network Hub Works

Imagine four computers connected to a four-port hub. If Computer A transmits an Ethernet frame, the hub repeats the signal to the ports connected to Computers B, C, and D.

The hub does not examine the frame's destination MAC address. It has no forwarding table and does not learn which address is connected to which port. Its operation is limited to restoring and retransmitting the signal.

Every connected endpoint receives the repeated physical transmission at its network interface. Each endpoint then examines the Ethernet frame. The endpoint whose MAC address matches the destination normally accepts the frame for Layer 2 processing. Other endpoints normally discard it.

Computer A  --\
Computer B  ---+-- [ Ethernet hub ]
Computer C  --/          |
Computer D  -------------+

A signal entering one port is repeated to the other active ports.

This physical repetition is different from a broadcast. A broadcast is a Layer 2 delivery method intended for all hosts in a broadcast domain. A hub repeats all incoming signals, including ordinary unicast traffic, without knowing whether the traffic is unicast, broadcast, or multicast.

OSI Model Placement: Layer 1

A hub operates at OSI Layer 1, the Physical layer. Layer 1 concerns the transmission of signals over a medium. Depending on the technology, those signals may be electrical, optical, or another form of physical encoding.

A hub does not interpret:

• Ethernet frame headers
• Source or destination MAC addresses
• IP addresses
• TCP or UDP port numbers
• Application protocols

An Ethernet frame is a Layer 2 unit of Ethernet data that contains source and destination MAC addresses. A MAC address is a Layer 2 hardware address assigned to a network interface. Although frames pass through a hub, the hub does not read or use these addresses.

A repeater is a Physical-layer device or function that restores and retransmits a signal. A hub applies this repeater behavior to multiple ports, making it a multiport Layer 1 device.

What Happens When a Hub Repeats a Frame?

Consider a frame sent from Host A to Host B:

1. Host A transmits a signal through its Ethernet interface.
2. The signal enters one port of the hub.
3. The hub regenerates the signal.
4. The hub sends the regenerated signal through every other active port.
5. Host B examines the destination MAC address and normally accepts the frame.
6. Hosts C and D also receive the physical transmission, but normally discard the frame because their MAC addresses do not match the destination.

The important distinction is that a hub repeats first and makes no destination decision. A switch can inspect the frame and often send a known unicast frame only toward the appropriate port.

Shared Bandwidth and Collision Domains

A collision domain is a network segment in which simultaneous transmissions can interfere with one another. All devices connected through a traditional hub belong to one shared collision domain.

The devices compete for the same shared medium and available bandwidth. If two hosts transmit at nearly the same time, their signals can overlap. The result is an Ethernet collision, which can corrupt the transmission.

Traditional shared Ethernet used CSMA/CD, or Carrier Sense Multiple Access with Collision Detection. In this system:

1. A host listens to determine whether the shared medium appears idle.
2. If the medium is idle, the host begins transmitting.
3. If another host transmits at the same time, a collision may occur.
4. The affected hosts stop, wait for a calculated backoff interval, and retry.

CSMA/CD is associated with shared, half-duplex Ethernet. It does not provide unlimited simultaneous use of the link. As more active devices share a hub, the probability of collisions increases and less of the link's capacity is available for useful data.

Example: A Collision on a Hub Segment

Suppose two computers connected to the same hub begin transmitting at nearly the same time. Because both use the same collision domain and half-duplex operation, their signals can interfere. The computers must stop and retry according to legacy Ethernet collision-handling behavior.

Frequent collisions cause retransmissions, delays, and reduced effective throughput. A lightly used hub may appear to work acceptably, but performance becomes less predictable as traffic and the number of active devices increase.

Half-Duplex Operation

Half-duplex communication allows transmission in both directions, but not at the same time on the shared segment. A device attached to a hub cannot reliably send and receive simultaneously because all devices share one medium.

Hubs operate in half-duplex mode. This limitation is closely related to collision handling: devices must coordinate access to the shared medium because simultaneous transmissions can interfere.

Full-duplex communication allows a device to send and receive simultaneously. Modern switch-to-host and switch-to-switch Ethernet links are typically point-to-point full-duplex connections. Since each link has its own dedicated path between two interfaces, collisions do not occur under normal operation.

Duplex settings matter when a legacy hub is involved. If a host interface is forced to full-duplex while the hub-side connection operates at half-duplex, the two ends do not use the medium in the same way. This is a duplex mismatch and can cause poor throughput, errors, and inconsistent communication.

Security and Traffic Visibility

Because a hub repeats traffic to all other ports, every connected host can physically receive transmissions sent by other hosts. A computer attached to an unused hub port can use a packet sniffer to capture and analyze frames visible to its interface without requiring switch port mirroring.

This creates a privacy and security disadvantage. On a hub-based segment, ordinary unicast traffic is not physically isolated to the destination port. A monitoring tool may be able to observe addressing information, protocol metadata, and application traffic.

Physical visibility does not automatically mean that the observer can read protected application data. Encryption, such as that used by properly configured secure protocols, can protect the content even when the traffic is captured. Encryption remains important on both shared and switched networks.

Hub Versus Ethernet Switch

A network switch is generally a Layer 2 device. It reads Ethernet frame information, learns source MAC addresses, and uses a MAC address table to associate addresses with ports.

When a switch receives a frame for a destination MAC address that it has learned, it normally forwards the frame only through the port associated with that address. If the destination is unknown, or if the frame is a broadcast, the switch may send it through multiple ports according to its forwarding rules.

A switch does not eliminate every form of traffic replication. For example, broadcasts are intended for all hosts in the relevant Layer 2 broadcast domain, and an unknown unicast may be temporarily flooded while the switch learns. However, a switch normally prevents known ordinary unicast traffic from being delivered to every port.

Performance and Operational Disadvantages

Hubs are generally unsuitable for modern production LANs because they provide no traffic isolation or intelligent forwarding. Their main disadvantages are:

• Increased collisions: All devices share one collision domain.
• Reduced effective throughput: Retransmissions and backoff consume capacity.
• Shared bandwidth: Devices compete for the same medium instead of using separate switched links.
• Half-duplex restrictions: Devices cannot reliably transmit and receive at the same time.
• Unnecessary signal propagation: Every incoming transmission is repeated to all other active ports.
• No traffic isolation: Ordinary unicast transmissions are physically visible at every connected interface.
• No intelligent forwarding: The hub has no MAC address table, VLAN configuration, or forwarding logic.

As more devices become active, the shared medium becomes busier. Collisions, waiting, and retransmissions increase, so users may experience slow or inconsistent communication.

Hub Configuration and Verification

A traditional unmanaged hub has no command-line interface and normally has no configuration for MAC addresses, VLANs, forwarding, or port security. It simply repeats signals as long as the physical link and signal characteristics are compatible.

Do not apply switch commands to a hub. A hub does not have a MAC address table to display or configure.

If the hub is replaced with a switch, the switch interfaces should use compatible speed and duplex settings. Modern devices often negotiate these settings automatically, but administrators may verify interface status and error counters when troubleshooting a switched replacement.

Troubleshooting Hub-Based Networks

Slow or Inconsistent Communication

Likely causes:

• High utilization of the shared segment
• Frequent collisions
• Too many active devices sharing the hub's bandwidth

Check collision-related interface statistics on attached devices or upstream equipment. Where possible, replace the hub with an Ethernet switch. This separates host links, removes the shared collision domain per port, and commonly enables full-duplex operation.

Duplex Mismatch or Poor Throughput

Likely causes:

• A device interface is forced to full-duplex while the hub connection operates at half-duplex
• Incompatible speed or duplex settings

Use a compatible half-duplex setting when a hub must temporarily be retained. Replacing the hub is the preferred remediation because it removes the shared-media limitation rather than merely accommodating it.

A Host Can Observe Other Hosts' Traffic

This is normally expected hub behavior, not a hub fault. The hub repeats transmissions from one port to the other active ports. Migrate the segment to a switch and use encryption for sensitive data.

Users Expect Destination-Port Forwarding

This symptom usually indicates confusion between a hub and a switch. Selective forwarding requires a Layer 2 switch with MAC address learning. A hub cannot direct traffic only to the destination port.

Why Hubs Still Matter

Hubs are largely obsolete in contemporary Ethernet LANs. Switches replaced them because switches provide better performance, scalability, full-duplex links, and traffic separation.

Even so, hubs remain useful as a conceptual topic. Understanding hubs helps explain:

• Why shared Ethernet has collision domains
• Why half-duplex operation requires access control
• How CSMA/CD relates to legacy Ethernet
• Why switches improve ordinary unicast forwarding
• Why traffic capture was easier on a shared physical segment
• Why duplex mismatches can damage performance

Key Takeaways

• A network hub is a legacy multiport Layer 1 repeater.
• It regenerates a signal received on one port and repeats it to every other active port.
• It does not inspect Ethernet frames, MAC addresses, IP addresses, or higher-layer protocols.
• All devices attached to a hub share one collision domain and shared bandwidth.
• Hubs operate in half-duplex mode, so simultaneous transmissions can collide.
• A switch learns MAC addresses and normally forwards known unicast frames only to the appropriate port.
• Hubs make physical traffic observation easier and provide poor traffic isolation.
• Modern LANs use switches instead of hubs, but hubs remain important for understanding legacy Ethernet behavior.

For the concise definition, see what is a network hub.