7.5	Effects of Layer 2 Devices on Data Flow
	7.5.1	Ethernet LAN segmentation
There are two primary reasons for segmenting a LAN. The first is to isolate traffic between segments, and to achieve more bandwidth per user by creating smaller collision domains. Without LAN segmentation, LANs larger than a small workgroup would quickly become clogged with traffic and collisions, and would deliver virtually no bandwidth. The addition of devices like bridges, switches, and routers segment the LAN (shown) into four collision domains. By dividing large networks into self-contained units, bridges and switches provide several advantages. A bridge, or switch, diminishes the traffic experienced by devices on all connected segments, because only a certain percentage of traffic is forwarded. Both devices act as a firewall for some potentially damaging network errors. They also accommodate communication between a larger number of devices than would be supported on any single LAN connected to the bridge. Bridges and switches extend the effective length of a LAN, permitting the attachment of distant stations that were not previously permitted. Although bridges and switches share most relevant attributes, several distinctions still do exist between them. Switches are significantly faster because they switch in hardware, while bridges switch in software, and can interconnect LANs of unlike bandwidth. A 10 Mbps Ethernet LAN and a 100 Mbps Ethernet LAN can be connected by using a switch. Switches can support higher port densities than bridges. Some switches support cut-through switching, which reduces latency and delays in the network, while bridges support only store-and-forward traffic switching. Finally, switches reduce collisions and increase bandwidth on network segments because they provide dedicated bandwidth to each network segment. Segmentation by routers has all of these advantages and more. Each interface on the router connects to a separate network, so insertion of the router into a LAN creates smaller collision domains and smaller broadcast domains. This occurs because routers do not forward broadcasts unless programmed to do so. However, the router can perform bridging and switching functions. The router can perform best path selection. The router can be used to connect different networking media, and different LAN technologies. Note the router in the teaching topology is connecting Ethernet, Token Ring and FDDI LAN technologies - segmenting the LAN, but doing much more. Routers can connect LANs running different protocols (IP vs. IPX vs. AppleTalk) and can have serial connections to WANs.

7.5	Effects of Layer 2 Devices on Data Flow
	7.5.2	Bridge segmentation of a collision domain
Ethernet LANs that use a bridge for segmenting the LAN provide more bandwidth per user because there are fewer users on the segments than there are when compared to the entire LAN. The bridge allows only those frames that have destinations outside the segment to pass through. Bridges learn a network’s segmentation by building address tables that contain the physical address of each network device, as well as the port to use to reach the device. Bridges differ from routers because they are Layer 2 devices, and are, therefore, independent of Layer 3 protocols. Bridges pass on data frames, regardless of which Layer 3 protocol is used, and are transparent to the other devices on the network. Bridges increase the latency (delay) in a network by 10-30%. This latency is due to the decision making that is required of the bridge, or bridges, when transmitting data to the correct segment. A bridge is considered a store-and-forward device because it must receive the entire frame and compute the cyclic redundancy check (CRC) before forwarding can take place. The time it takes to perform these tasks can slow network transmissions, thus causing delay.

7.5	Effects of Layer 2 Devices on Data Flow
	7.5.3	Switch segmentation of a collision domain
A LAN that uses a switched Ethernet topology creates a network that performs as though it had only two nodes – the sending node and the receiving node. These two nodes share 10 Mbps bandwidth between them, which means nearly all bandwidth is available for the transmission of data. A switched Ethernet LAN allows a LAN topology to work faster and more efficiently than a standard Ethernet LAN can, because it uses bandwidth so efficiently. In a switched Ethernet implementation, the available bandwidth can reach close to 100%. It is important to note that even though 100% of the bandwidth may be available, Ethernet networks perform best when kept under 30-40% of full capacity. This limitation is due to Ethernet’s media access method (CSMA/CD). Bandwidth usage that exceeds the recommended limitation results in increased collisions. The purpose of LAN switching is to ease bandwidth shortages and network bottlenecks, such as that occurring between a group of PCs and a remote file server. A LAN switch is a high-speed multi-port bridge that has one port for each node, or segment, of the LAN. A switch segments a LAN into micro-segments, thereby creating collision free domains from one formerly larger collision domain. Switched Ethernet is based on standard Ethernet. Each node is directly connected to one of its ports, or to a segment that is connected to one of the switch's ports. This creates a 10 Mbps connection between each node and each segment on the switch. A computer connected directly to an Ethernet switch is its own collision domain and accesses the full 10Mbps. As a frame enters a switch it is read for the source and/or destination address. The switch then determines which switching action will take place based on what is learned from the information in the frame. If the destination address in located on another segment, the frame is then switched to its destination.

7.5	Effects of Layer 2 Devices on Data Flow
	7.5.4	Router segmentation of a collision domain
Routers are more advanced than typical bridges. A bridge is passive (transparent) at the network layer and operates at the data link layer. A router operates at the network layer, and bases all of its forwarding decisions on the Layer 3 protocol address. It accomplishes this by examining the destination address on the data packet, then looking in its routing table for forwarding instructions. Routers create the highest level of segmentation because of their ability to make exact determinations of where to send the data packet. Because routers perform more functions than bridges, they operate with a higher rate of latency. Routers must examine packets to determine the best path for forwarding them to their destinations. Unavoidably, this process takes time and introduces latency.

7.5	Effects of Layer 2 Devices on Data Flow
	7.5.5	Teaching topology segmentation by bridges, switches, and routers
The teaching topology contains examples of segmentation by bridges, switches, and routers. Also in the teaching topology, many different parts of the network are brought together by the main router. The bridge divides the E1 Ethernet network into two segments. Traffic is filtered at the bridge, reducing potential collisions and the physical extent of the collision domain. Therefore, the bridge breaks the E1 Ethernet network into two segments: the first segment has the repeater and hosts K, L, M, N on it; the second segment has hosts O and P on it. This remains, however, a broadcast domain. The repeater extends the collision domain rather than segmenting it. The main switch divides the E0 Ethernet network into multiple network segments with each having guaranteed full bandwidth. The workgroup switch divides the workgroup segment into more segments. There are no broadcast domains on the segments off of the main switch or the workgroup switch. Also note that the switches provide high connectivity to their unshared bandwidth. The hub does not segment its part of the network. The hub and all the devices attached to it, all the way up to the main switch port, remain a collision domain. The router segments the entire LAN into two Ethernet subnetworks, which are segmented, and a Token-Ring and FDDI subnetwork, which by their nature, have no collision domains.