COMPUTER NETWORKING:THE NETWORKING INFRASTRUCTURE

THE NETWORKING INFRASTRUCTURE

As mentioned in Section 2.1, the background of the information transfer is provided by the net- working infrastructure. As mentioned, the solution of the problem of how to handle millions of transfers of data simultaneously over the global network in very short time frames is based on a hierarchical model. The infrastructure consists of links and nodes whose role depends on their status in the hierarchy. At the highest level are the largest information exchange centers, which are connected with very high-speed lines. The very high-speed networks of these large exchange nodes and the high-capacity links connecting them are called backbones. The main nodes of the backbones com- municate among each other and also collect and distribute the relevant traffic from and to the regional nodes around them. The lines connecting the large exchange nodes to the regional nodes also work at very high speeds and are therefore able to transmit very high-volume traffic.

Lower in the hierarchy, around the regional nodes, somewhat lower-speed access lines transfer the information from and to the area servers (connected to routers and / or switches, i.e., special machines taking care of the direction of the transmitted information packets in the network) within the region. These active devices serve either LANs, MANs, or WANs, depending on the topology and the further hierarchical structure of the network around these area servers. Their task is to collect/ distribute the information traffic in their area.

The hierarchical structure might mean a simple (but of course highly segmented and in this respect very complex) tree-like topology. However, as already mentioned, for the purpose of reliability and routing efficiency, cross-connections (redundant routes) are also established within the infrastructure. Obviously, the difficulty of handling the traffic is in relation with the complexity of the network structure, but this is the price for satisfactory global network operation.

The key elements, besides the communication links themselves, in this technique of information transfer through the network are the routers and switches, which take care of directing the information packets through the nodes in this hierarchy so that they finally arrive at their destination.

The speed of the information transfer depends on the bandwidth of the links and the processing speed of the routers and switches in the infrastructure. Taking into account the requirements stemming from the characteristics of the transmitted information and the traffic volume, that is, from the esti- mated average and peak number of requested transmission transactions, allows the speed of the different sections of the infrastructure to be determined. Some basic figures are as follows:

• Simple alphanumeric messages require a speed that is not too high at the last mile sections, that is, at the lines closest to the transmitters / receivers of the information. Several Kbps is considered satisfactory in such cases of message delivery without special requirements regarding the delivery time.

• Real-time transmission of similarly simple alphanumeric information requires either similar speed, as above, or, in case of higher volume of these kinds of data, a certain multiple of that speed.

• Real-time transmission of sampled, digitized, coded, and possibly compressed information, stemming from high-fidelity voice, high-resolution still video, or high-quality real video signals do require much higher speed. The last-mile speed in these cases ranges from dozens of Kbps to several and even hundreds of Mbps (HDTV). This information is quite often cached by the receiver to overcome short interruptions in the network’s delivery of the information at high speed, providing the receiver a smooth uninterrupted flow of information.

This means that although until recently the last-mile line speeds were in the range of several dozens of Kbps, the area access lines were in the range of several Mbps, and the top-level backbones were approaching the Gbps speed, lately, during the 1990s, these figures have increased to Mbps, several hundreds of Mbps, and several Gbps level, respectively. The near future will be characterized by multi-Gbps speed at the backbone and area access level, and the goal is to reach several hundreds of Mbps even at the last-mile sections. Of course, the development of the infrastructure means coexistence of these leading-edge figures with the more established lower-speed solutions. And ob- viously, in the longer term, the figures may increase further.

As far as the organization of the operation of the infrastructure, the physical and the network infrastructure should be distinguished. Providing physical infrastructure means little more than mak- ing available the copper cables or fibers and the basic active devices of the transmission network, while operating the network infrastructure also means managing of the data traffic. Both organiza- tional levels of the infrastructure are equally important, and the complexity of the related tasks has, in an increasing number of cases, resulted in the jobs being shared between large companies spe- cializing in providing either the physical or the network infrastructure.

The first step from physical connectivity to network operation is made with the introduction of network protocols. These protocols are special complex software systems establishing and controlling appropriate network operation. The most important and best-known such protocol is the Internet Protocol.

The elementary services are built up on top of the infrastructure outlined above. Thus, the next group of companies working for the benefit of the end users is the Internet service providers (ISPs), which take care of providing network access points.

Although network services are discussed separately in Section 7, it should be mentioned here that one of the basic tasks of these services is to take care of the global addressing system. Unique addressing is perhaps the most important element in the operation of the global network. Network addresses are associated with all nodes in the global network so that the destination of the transmitted information can always be specified. Without such a unique, explicit, and unambiguous addressing system, it would not be possible even to reach a target site through the network. That is why ad- dressing (as well as naming, i.e., associating unique symbolic alphanumeric names with the numeric addresses) is a crucial component of network operation. The task is solved again by a hierarchy of services performed by the domain name service (DNS) providers. This issue is dealt with in more detail later. Note that it is possible, as a common practice, to proxy the end-node computer through an access point to share the use of a unique address on the Internet. The internal subnetwork address may in fact be used by some other similarly isolated node in a distinctly separate subnetwork (separate intranet).

Everything briefly described above relates to the global public network. However, with the more sophisticated, serious, and sometimes extremely sensitive usage of the network, the need to establish closed subsets of the networked computers has emerged. Although virtual private networks (subnet- works that are based on public services but that keep traffic separated by the use of special hardware and software solutions) solve the task by simply separating the general traffic and the traffic within a closed community, some requirements, especially those related to security, can only be met by more strictly separating the related traffic. The need for this kind of separation has led to the estab- lishment of intranets (special network segments devoted to a dedicated user community, most often a company) and extranets (bunches of geographically distant but organizationally and / or coopera- tively connected intranets, using the public network to connect the intranets but exploiting special techniques for keeping the required security guarantees). Although building exclusively private net- works, even wide area ones, is possible, these are gradually disappearing, and relying on public services is becoming common practice even for large organizations.