Recognizing A Network Based On Protocols

Description

Types of local networks

Main Article Text

We will take up and then conclude our discussion regarding the various types of local area networks. In the last issue we had seen the different families of LAN grouped by topology (the way in which the actual physical electric connection is made between the various stations), now it is time to examine the other side of the coin: the protocols. Almost all of the local networks use the method of subdividing the information into packets of bits, which are then transferred from one station to another through mechanisms identified by the name of Protocol. There are two basic types: deterministic and probabilistic. The first one collects all of the protocols that regulate access to the network using a system of variance: the transmission medium is reserved for the first station that takes possession. The second category includes arbitrament technologies, where there is a mechanism that establishes access, in rounds, for which workstations are empowered to transmit. Each of the two approaches has strengths and weaknesses. In particular the probabilistic method is more flexible and is well suited to manipulate quickly messages of medium size. However, if the network must support a strong workload and it is essential to ensure that each workstation has the opportunity to speak, or one must also ensure that the transmission is not stopped halfway, to move towards the deterministic technique is better. We will begin to identify the immediate families of the various networks available on the market, starting from the probabilistic camp. Here we find Ethernet, AppleTalk, StarLan, 10-Net, Corvus OmniNet, G Net, IBM PC Network and others. The most widespread commercial version of the probabilistic protocol is called CSMA (Carrier sense multiple access – multiple access with carrier recognition), which in itself is divided into two versions CSMA CD (Collision Detection – identification and collision detection) and Ca CSMA (Collision Avoidance – avoidance of collisions). The CSMA Ca, the older of the two versions,is now falling into disuse. Following this protocol the workstation listens to see if the network is free before proceeding to transmit, and then sends its message. The method is , however, ineffective because it does not guarantee the success of the link. There is, in fact, the possibility that two nodes, after having ascertained that the path is clear, begin transmitting at the same time so that the messages collide. Unfortunately, lacking the Cd function (Automatic collision detection) both stations continue to transmit and only when they discover that they haven’t received the confirmation receipt by the consignee do they notice that something has gone wrong. The CSMA Cd, therefore, is preferable to CSMA Ca because it immediately stops transmission in the event of a conflict between the packages, because in this conflict, it prevents both stations involved from transmitting simultaneously, which would cause the possibility of a second collision. The waiting time to which the node is subject after the collision increases according to the load present in LAN.

The deterministic protocol finds practical applications in two products, the IBM’s Tokenring network, and the ArcNet of Datapoint network. Token-passing is the trade name the protocol assumes in both cases, but there are operational differences in both environments. In the case of Ibm, the electrical topology and the structure of communications are closely linked. Let us recall that the token-ring is , in fact, a ring from the electrical point of view and that the signal travels from one station to another outlining g an imaginary circle. The passage of the token (witness) must, therefore ,adapt to this scheme. The token is a packet of bits that rotate along the ring at a very high speed, but can be intercepted by any of the connected nodes which then changes the state from free to busy and while claiming the message he wants to convey places it in the circle. At that point, the witness continues its route and may only be detained by the station where the communication is intended. The latter then collects the package’s incoming information ensuring that the token continues on its route together with confirmation of receipt. The tour is complete when the transmitting station receives the token back and can now decides to release it, or use it to transmit a second packet. This procedure thus prevents the transmission of complexes packets (such as images, videos, or music) but can be interrupted by unexpected collisions or conflicts between one station and another. In addition, under heavy traffic conditions of, there are no traffic jams due to the continued interference among users who try to speak simultaneously. The token-passing mechanism used on ArcNec networks is , however, a special type, as it is an electrical form of this network and is a bus with the physical topology of a distributed star (i.e. There are several concentrated devices from which radiating coaxial cables depart that connect the individual nodes while being a concentrator to the other horizontal connection line). As the network has a linear configuration (the two extremities do not converge to form a circle), the circulation path of the bus will be a logical ring and not a physical one. This means that one has to tell the token which way to go, indicating where it has to go, to each station address. In June 1982, the International Electrical and Electronics Engineers (IEEE) published studies showing the performance obtained with these three different protocols. Without going into details of the bit and frequencies, we can say that the document shows that the CSMA CD offers excellent performance when the amount of stations networked are quite low, but tends to lose the efficiency when there are a hundred users working at the same time. Token-ring and token-bus, however, react favorably to an overload, showing no appreciable reduction in the response time of the network.

Roberto Mazzoni