[Contents] [Back] [<< Prev] [Next >>]

Bridges
Bridges, on the other hand, can extend the maximum size of a network. Although the bridged network in Figure 6.6 looks much like the earlier example of a network with a repeater, the bridge is a much more flexible device. Bridges operate at the MAC sublayer of the OSI Data Link layer (see Chapter 2).

A repeater passes on all signals that it receives. A bridge, on the other hand, is more selective and passes only those signals targeted for a computer on the other side. A bridge can make this determination because each device on the network is identified by a unique address. Each packet that is transmitted bears the address of the device to which it should be delivered. The process works as follows:
1.The bridge receives every packet on LAN A and LAN B.
2.The bridge learns from the packets which device addresses are located on LAN A and which are on LAN B. The bridge then builds a table with this information.
3.Packets on LAN A that are addressed to devices on LAN A are discarded, as are packets on LAN B that are addressed to devices on LAN B. These packets can be delivered without the help of the bridge.
4.Packets on LAN A addressed to devices on LAN B are retransmitted to LAN B for delivery. Similarly, the appropriate packets on LAN B are retransmitted to LAN A.
On older bridges, the network administrator had to manually configure the address tables. Newer bridges are called learning bridges. Learning bridges function as described in step 2, automatically updating their address tables as devices are added to or removed from the network.

Bridges accomplish several things. First, they divide busy networks into smaller segments. If the network is designed so that most packets can be delivered without crossing a bridge, traffic on the individual network segments can be reduced. If the Accounting and Sales departments are overloading the LAN, for example, you might divide the network so that Accounting is on one segment and Sales on another. Only when Accounting and Sales must exchange packets does a packet need to cross the bridge between the segments.

Bridges also can extend the physical size of a network. Although the individual segments still are restricted by the maximum size imposed by the network design limits, bridges enable network designers to stretch the distances between segments and extend the overall size of the network.

Bridges, however, cannot join dissimilar types of LANs. This is because bridges depend on the physical addresses of devices. Physical device addresses are functions of the Data Link layer, and different Data Link layer protocols are used for each type of network. A bridge, therefore, cannot be used to join an Ethernet segment to a Token Ring segment.

Bridges sometimes are also used to link a LAN segment through a synchronous modem connection to another LAN segment at a remote location. A so-called remote bridge minimizes modem traffic by filtering signals that won’t need to cross the modem line (see Figure 6.7).