There are two modes to send Ethernet frames. These modes are full-duplex and half-duplex. In the full-duplex mode, a device can send and receive frames simultaneously. In the half-duplex mode, a device can either send or receive frames at a given time.
Full-duplex and half-duplex are device-specific. A device such as a hub that understands only signals can work only in half-duplex mode. A device such as an Ethernet switch that can convert signals into frames to make forwarding decisions can work in both modes.
Let's take an example to understand both modes.
Half-duplex example
Suppose three PCs are connected in a LAN through a Hub. A hub forwards data using physical layer standards rather than data-link layer standards. When an electrical signal enters one hub port, the hub repeats that electrical signal out of all other ports (except the incoming port).
The signal reaches all the rest of the devices connected to the hub. The destination device converts signals into frames and processes them. The hub has no concept of Ethernet frames.
If two or more devices transmit signals at the same time, the electrical signals collide and become useless for the destination device. The destination device cannot read garbled signals.
Half-duplex mode solves this problem. In half-duplex mode, a device checks the wire before placing its signals on the wire. If the wire contains signals, the device assumes another device is transmitting signals. It waits until another device's transmission is over. It places its signals on the wire only when the wire has no signals.
Full-duplex example
Let's use the preceding example again. But this time, replace the hub with a switch. A switch uses data link-layer standards to forward data. When electrical signals enter one switch port, it converts the signals into frames, reads destination addresses in the frames, and forwards each frame only from the port that is connected to the destination address.
If multiple frames have the same destination, it arranges them in a queue before forwarding them to the destination.
Differences between Half-duplex and full-duplex
The following table lists the differences between half-duplex and full-duplex.
| Chararsticis | Half-duplex | Full-duplex |
| Data transmission | Send or receive at the same time | Send and receive at the same time |
| Device | Hub, bridge | Ethernet Switches |
| Collisions | Yes | No |
| CSMA/CD | Enabled | Disabled |
| Data transfer speed | Slow | Fast |
| Waiting time | The device must wait to send if another device is transmitting. | No waiting time. A device can send data at any time. |
Duplex on Cisco switches
Nowadays, hubs and bridges are no longer used in computer networks. Almost all modern computer networks use Ethernet switches. For backward compatibility, Cisco switches support both modes on per-port basis. The default mode is full duplex. If your network has only switches, you should run all switches on full duplex mode. If any switch port is connected to a hub or a bridge, the switch automatically changes duplex mode on that port to half-duplex.
