Switches are enhanced versions of bridges . Bridges were introduced in the mid-1980s to improve network performance. They solved a basic network problem: reducing network collisions by segmenting networks . On an Ethernet segment, only one machine may transmit at once. If more than one machine tries to communicate, a collision occurs on the segment. When a collision occurs, the machines that were trying to communicate go into a random wait period before attempting to transmit again. As the number of devices on a segment increases, so does the number of collisions. And the more collisions on a segment, the worse the network performs. By using a bridge, a network is separated into segments called collision domains , which reduce the number of devices—and collisions—per segment.
Switches improve on bridges in one important way; switches allow the network to be partitioned into logical smaller segments called Virtual LANs or VLANs. These VLANs allow you to create even smaller domains, decreasing the likelihood of collisions and improving network performance.
Bridges and switches act almost the same way when it comes to learning MAC addresses and forwarding packets based on those addresses. Both switches and bridges implement a loop prevention protocol called spanning tree, described later in this chapter.
When switches were first introduced, the companies selling switches announced the death of the router! We now know that switches and routers must work ...