Chapter 3

How Bridges Work


CONTENTS

An intranet-especially one in a large corporation-can be a remarkably complex endeavor, consisting of many networks, including departmental local area networks, as well as larger subnetworks, which each are in essence collections of networks.

Bridges connect networks. When bridges were first developed they could not connect incompatible networking architectures. Early bridges functioned at the media-access control (MAC) section of the data-link layer of the OSI model (discussed in Chapter 2. Ethernet and Token Ring architecture differ at the MAC section of the data-link layer, hence their incompatibility. Newer bridges function at the logical-link control (LLC) portion of the data-link layer. As long as the communication protocol on both networks is the same (IPX to IPX, for example), Ethernet and Token Ring networks can be linked by bridges. Connecting divergent networks is an essential requirement for intranets.

Bridges are also used to connect networks indirectly by long-distance, usually leased, lines. Remote bridges on Ethernet networks use a transparent routing technique to handle traffic. It is the destination address of each packet that is read to determine the action to be taken. If the destination is on the same network, bridges ignore the packet; bridges pass only packets that need to go to other networks. Token Ring operates somewhat differently in that it uses source routing. In this case, the bridge uses a test message to calculate the best path between the source and the destination.

In order to determine whether to pass or drop a packet, bridges refer to a table, which is basically a list of addresses. These tables originally had to be built manually, causing considerable maintenance every time a replacement or additional NIC address was put on the network. This led to the development of learning bridges, that is, bridges with software algorithms that could build these tables automatically. Today's bridges build and maintain their tables by listening to cable traffic and checking packets for source addresses.

Bridges are relatively simple to set up and maintain, and they can operate fairly fast since they simply decide whether to pass or drop the packet. In the next chapter a similar but more complex piece of equipment, the router, is discussed. Bridges are often used to segment and reduce local network traffic. Routers are usually used for connecting entire networks and subnetworks together.

Bridges and routers are sometimes combined into a single product called a brouter. A brouter combines the functions of both bridge and router. It examines the outermost address to see if the data needs to be sent to another LAN over a bridge, and delivers information that way if possible. If, instead, it needs to be sent using IP technology, it will act as a router does.

How Bridges Work

Bridges are hardware and software combinations that connect different parts of a single network, such as different sections of an intranet. They connect local area networks (LANs) to each other. They are generally not used, however, for connecting entire networks to each other, for example, for connecting an intranet to the Internet, or an intranet to an intranet, or to connect an entire subnetwork to an entire subnetwork. To do that, more sophisticated pieces of technology called routers are used. (See Chapter 4for more information on how routers work.)

  1. When there is a great amount of traffic on an Ethernet local area network, packets can collide with one another, reducing the efficiency of the network, and slowing down network traffic. Packets can collide because so much of the traffic is routed among all the workstations on the network.
  2. In order to cut down on the collision rate, a single LAN can be subdivided into two or more LANs. For example, a single LAN can be subdivided into several departmental LANs. Most of the traffic in each departmental LAN stays within the department LAN, and so it needn't travel through all the workstations on all the LANs on the network. In this way, collisions are reduced. Bridges are used to link the LANs. The only traffic that needs to travel across bridges is traffic bound for another LAN. Any traffic within the LAN need not travel across a bridge.
  3. Each packet of data on an intranet has more information in it than just the IP information. It also includes addressing information required for other underlying network architecture, such as for an Ethernet network. Bridges look at this outer network addressing information and deliver the packet to the proper address on a LAN
  4. Bridges consult a learning table that has the addresses of all the network nodes in it. If a bridge finds that a packet belongs on its own LAN, it keeps the packet inside the LAN. If it finds that the workstation is on another LAN, it forwards the packet. The bridge constantly updates the l
    earning table as it monitors and routes traffic.
  5. Bridges can connect LANs in a variety of different ways. They can connect LANs using serial connections over traditional phone lines and modems, over ISDN lines, and over direct cable connections. CSU/DSU units are used to connect bridges to telephone lines for remote connectivity.
  6. Bridges and routers are sometimes combined into a single product called a brouter. A brouter handles both bridging and routing tasks. If the data needs to be sent only to another LAN on the network or subnetwork, it will act only as a bridge delivering the data based on the Ethernet address. If the destination is another network entirely, it will act as a router, examining the IP packets and routing the data based on the IP address.