by Dinesh Thakur Category: Switching

The Ethernet switch or Ethernet FDSE (Full Duplex Switched Ethernet), was born in the early 1990s before the advent of switched Ethernet, shared Ethernet networks were cut into shared subnets autonomous, interconnected by bridges. Therefore, the traffic was multiplied by the number of subnets.

The bridges are in fact only Ethernet switches that store frames and retransmit them to other Ethernet networks. Continuing this logic to the extreme, we can cut the network to have only one station by Ethernet. Ethernet switching is then obtained.

The Ethernet network is a particularly simple FDSE network since there are only two stations: that we want to connect to the network and the connection switch. Therefore has a terminal directly connected by Ethernet to the switch.

In the Ethernet switching, each card coupler is directly connected to a switch Ethernet, which is responsible for redirecting the frames in the right direction. Must to talk about here since a frame preamble makes it possible to detect the start. Switching request a reference which, a priori, does not exist in the Ethernet world, no supervisory packet opening the virtual circuit, asking references. The switch password may be considered as inaccurate since there is no reference. However, it is possible to talk about switching if the recipient's address is regarded as a reference. The virtual circuit is then determined by the following references equal to the destination address of 6 bytes. To temporarily switch from end to end, each switch must be able to determine the output link according to the reference, that is to say the value of the receiver's address.

This switching technique may have the following problems:

• Address management of all couplers connected to the network.

• Management of possible congestion in a switch.

Due to the second difficulty, we must implement control techniques to take charge, on routes between switches, the frames simultaneously from all Ethernet modules. This echoes the characteristics of switching networks architectures. The technique of recovery on a collision is no longer useful since there is no collision in this case. In contrast, no distance limitation exists more, we can achieve Ethernet switching network to the size of the planet.

The Ethernet environment is now required by its implementation simplicity as the network remains limited in size. It is a networking solution that has the advantage of relying on the existing, namely couplers and various Ethernet networks that many companies have implemented to create their local networks.

Since the data produced in IP format, IPX (Internetwork Packet eXchange) or both are present in Ethernet frames in order to be carried in the local environment; it is tempting to switch directly to this structure on routes between LANs. Like all networks are compatible Ethernet environment, all machines emitting Ethernet frames speak the same language and can easily interconnect. We can perform extremely complex networks, with segments divided on the local parties and switched connections over long distances or between Ethernet switches.

The disadvantage of the switching frame level lies in the addressing of level 2, which corresponds to the flat Ethernet addressing. The dish, or absolute addressing, does not know the geographical location of a coupler according to its value. Once the network has a large number of positions, which is the case if one accepts a terminal mobility, for example, updating routing tables becomes almost impossible as there is no standardization for automation of this function.

The limitation of the performance of the Ethernet environment is due to sharing physical media by all couplers. To remedy this drawback, can increase the base rate by switching to 100 Mbit/s or 1 Gbit/s. Another solution is to switch the Ethernet frames. The first step towards switching consists, as we have seen, Ethernet networks to cut into small sections and connect them by a bridge. The role of the bridge is to filter frames allowing only those to an Ethernet network other than that from which the frame. Therefore, we limit the number of couplers that share the same Ethernet network. For this solution to be viable, the traffic must be relatively local.

In switching, Ethernet is reduced to its simplest form: a coupler Ethernet. So there is more of a collision. Next address, switch routes the frame to another switch or to a coupler. The available capacity per terminal is 10 Mbit/s, 100 Mbit/s or 1 Gbit / s. The difficulty lays in the complexity of Ethernet frames switched networks, with problems opening paths and flow control they pose.

Since there are a lot of Ethernet modules worldwide, the switched Ethernet the solution is booming. The Ethernet modules should be simplified, the power of the access technology is no longer necessary in this case.

As explained, one of the great advantages of this technique is no longer present distance restriction, since we switched mode. The distances between connected machines can reach several thousand kilometers. Another advantage is provided by the increased throughput per terminal. The transmission capacity can reach 10, 100 Mbit/s or 1 Gbit/s per machine. The counterpart of this flow gain is the return to a switched mode, where you have to draw paths for switching, implement flow control and complete management of physical addresses of the couplers. In other words, each switch needs to know the MAC address of all couplers connected to the network and know which direction to send the frames.

In companies, the Ethernet network can be a combination of shared networks and switched networks, Ethernet networks are all compatible to the frame level. If the corporate network is too vast to allow management of all the addresses in each switch, divide the network into separate domains. Two solutions are possible.

• Use the concept of virtual private network.

• Switching from one domain to another, going back to the network architecture level (Layer 3 of the OSI architecture), that is to say, by recovering information conveyed in the data frame area and by using the network level address to perform the routing. This transfer element is none other, of course, a router.

There are two main types of switching:

• The switch port, wherein the couplers are directly connected to the switch.

• The switching segment, wherein they are shared Ethernet network segments which are interconnected.

These two solutions are illustrated in Figure. The upper part of Figure shows a switch in which each port is connected to one terminal station. In the lower part of the figure, the switch has two ports on each which shared Ethernet is connected.

  Both types of switching

About Dinesh Thakur

Dinesh ThakurDinesh Thakur holds an B.C.A, MCSE, MCDBA, CCNA, CCNP, A+, SCJP certifications. Dinesh authors the hugely popular blog. Where he writes how-to guides around Computer fundamental , computer software, Computer programming, and web apps. For any type of query or something that you think is missing, please feel free to Contact us.