Network Bridging

Packet forwarding generally uses two processes. Of these two, bridging does not depend on address locations in a network, unlike routing; instead, bridging uses flooding, in which every incoming packet is sent through every possible link, and looks at the packet header source addresses to find devices, then records the MAC address so that it will not need to be tracked down again. This process severely limits the uses of bridging, so it is typically only used for local area networks. Additionally, flooding can be expensive and often results in duplicate packets.

The term 'bridging' is most often used to refer to 'transparent bridging' or 'learning bridge operation'. Token ring networks use source route bridging, an alternate form developed specifically for that purpose. In source route bridging, networks segments are connected at the DLL of the OSI model (layer 2).

Formally speaking, when referring to ethernet networks, a 'bridge' is an IEEE 802.1D compliant device. A 'switch' is a bridge with multiple ports, and the terms 'switch' and 'layer 2 switch' can and are frequently used as synonyms for bridge.

Repeaters and network hubs, which connect segments at the PHY of the OSI model (layer 1), share much in common with bridges but are much less complex. Bridges manage one-network traffic instead of rebroadcasting it to other network segments, and analyze incoming data packets to ensure that the bridge can send them on to other parts of the network.