Collisions are part of the operation of an Ethernet network because many device share the same physical segment a collision occurs when two or more hosts attempt to communicate at the same time, thus they are said to be in the same collision domain. This also implies that hosts will only get a percentage of available bandwidth due to shared connectivity.
Bridges allow an Ethernet network to be segmented into multiple collision domains thereby reducing the number of collisions and increasing the available bandwidth.
As shown in the picture above, a Bridge segments an Ethernet network into two collision domains(CCNA exam question). By doing so each host will share the bandwidth only with hosts in the same segment (collision domain). To enable communication between the segments a bridge has to a little more work. The 802.1D specification for transparent bridging defines some functions the bridge has to perform in order to transfer frames from one segment to the other. These functions are mentioned below
- Learning: It’s a process of learning a MAC address of a system and associating it with a port. A Bridge creates a table called content addressable memory (CAM) table to store this information.
- Flooding: A Bridge will flood a frame out all ports except the port on which it is received when it does not have an entry for the frame destination in the CAM table
- Filtering: After a Bridge has built a CAM table it will filter frames from going on to other segments when the frame destination is on the same segment as the frame source. In Figure 1, traffic between Host 1 and 2 will remain on the left segment and it will be filtered by bridge preventing it from being forwarded on the right segment.
- Forwarding: Its process of forwarding frames between the segments. In Figure 1 the Bridge will forward traffic between the segments if Host 1 needs to communicate with Host 2.
- Aging: After a Bridge learns a MAC Address on a port it will start an aging timer, if the Bridge does not hear the same MAC Address before the expiration of the aging timer the bridge will remove the entry from the CAM table.
Like bridges, switches perform the same function of segmenting the network into multiple collision domains, but they have many ports with each port usually connecting a single host and hence resulting in micro segmentation of the Ethernet network where each port is in a different collision domain. A switched Ethernet network replaces shared Ethernet network increasing network bandwidth because each switch port has access to full bandwidth.
Figure 2 shows a switched Ethernet network. Each host is in a different collision domain thereby increasing the bandwidth of each port to full bandwidth capacity.
A Cisco Switch can forward frames using three different switching methods mentioned below and covered on the CCNA 640-802 exam:
- Store and Forward: Using this method a switch copies each frame completely into the switch memory and computes a cyclic redundancy check (CRC) for errors. If any errors are found the frame is discarded. This method ensures error-free network traffic but the latency (delay) is increased due to the inspection of each frame and completely copying the frame into the memory before inspection. This also implies more memory and cpu power to carry out the processing as compared to other methods
- Cut-Through: The switch only copies the destination MAC address which is located in the first six bytes of the frame and forwards the frame immediately after it finds an outgoing port for the MAC address in the CAM table. Unlike Store and Forward switching method the switch will not check any frames for errors and as a result, corrupt frames are also forwarded. Although this method reduces latency inside the switch since switch forwards the corrupt frames to hosts, those frames would need to be resent by the source when the destination finds them being corrupted, and as a result, bandwidth is wasted.
- Fragment-Free: This method is a hybrid of the two other switching methods. It provides low latency error-free switching. The switch will copy only the first 64-bytes of the frame and check for errors caused due to collision before forwarding it. Frames corrupted due to collision can be identified within 64-bytes of frames thus the switch only needs to copy the first 64-bytes in the memory.
Today we covered a few important concepts related to bridges and switches. It is very important to properly understand these topics as they are the foundation for many other lessons in the Cisco CCNA Certification.