NETWORKING CONCEPTS

The Way of a WAN

To at least some extent, WANs are defined by their methods of transmitting data packets. True, the means of communication must be in place. True, too, the networks making up the WAN must be up and running. And the administrators of the network must be able to monitor traffic, plan for growth, and alleviate bottlenecks. But in the end, part of what makes a WAN a WAN is its ability to ship packets of data from one place to another, over whatever infrastructure is in place. It is up to the WAN to move those packets quickly and without error, delivering them and the data they contain in exactly the same condition they left the sender, even if they must pass through numerous intervening networks to reach their destination.
Picture, for a moment, a large network with many subnetworks, each of which has many individual users. To the users, this large network is (or should be) transparent—so smoothly functioning that it is invisible. After all, they neither know nor care whether the information they need is on server A or server B, whether the person with whom they want to communicate is in city X or city Y, or whether the underlying network runs this protocol or that one. They know only that they want the network to work, and that they want their information needs satisfied accurately, efficiently, and as quickly as possible.
Now picture the same situation from the network's point of view. It "sees" hundreds, thousands, and possibly even tens of thousands of network computers or terminals and myriad servers of all kinds—print, file, mail, and even servers offering Internet access—not to mention different types of computers, gateways, routers, and communications devices. In theory, any one of these devices could communicate with, or transmit information through, any other device. Any PC, for instance, could decide to access any of the servers on the network, no matter whether that server is in the same building or in an office in another country. To complicate matters even more, two PCs might try to access the same server, and even the same resource, at the same time. And of course, the chance that only one node anywhere on the network is active at any given time is minuscule, even in the coldest, darkest hours of the night.
So, in both theory and practice, this widespread network ends up interconnecting thousands or hundreds of thousands of individual network "dots," connecting them temporarily but on demand. How can it go about the business of shuffling data ranging from quick e-mails to large (in terms of bytes) documents and even larger graphic images, sound files, and so on, when the possible interconnections between and among nodes would make a bowl of spaghetti look well organized by comparison? The solution is in the routing, which involves several different switching technologies.
Switching of any type involves moving something through a series of intermediate steps, or segments, rather than moving it directly from start point to end point. Trains, for example, can be switched from track to track, rather than run on a single, uninterrupted piece of track, and still reach their intended destination. Switching in networks works in somewhat the same way: Instead of relying on a permanent connection between source and destination, network switching relies on series of temporary connections that relay messages from station to station. Switching serves the same purpose as the direct connection, but it uses transmission resources more efficiently.
WANs (and LANs, including Ethernet and Token Ring) rely primarily on packet switching, but they also make use of circuit switching, message switching, and the relatively recent, high-speed packet-switching technology known as cell relay.

Top of page

Circuit Switching

Circuit switching involves creating a direct physical connection between sender and receiver, a connection that lasts as long as the two parties need to communicate. In order for this to happen, of course, the connection must be set up before any communication can occur. Once the connection is made, however, the sender and receiver can count on "owning" the bandwidth allotted to them for as long as they remain connected.
Although both the sender and receiver must abide by the same data transfer speed, circuit switching does allow for a fixed (and rapid) rate of transmission. The primary drawback to circuit switching is the fact that any unused bandwidth remains exactly that: unused. Because the connection is reserved only for the two communicating parties, that unused bandwidth cannot be "borrowed" for any other transmission.
The most common form of circuit switching happens in that most familiar of networks, the telephone system, but circuit switching is also used in some networks. Currently available ISDN lines, also known as narrowband ISDN, and the form of T1 known as switched T1 are both examples of circuit-switched communications technologies.

Top of page

Message Switching

Unlike circuit switching, message switching does not involve a direct physical connection between sender and receiver. When a network relies on message switching, the sender can fire off a transmission—after addressing it appropriately—whenever it wants. That message is then routed through intermediate stations or, possibly, to a central network computer. Along the way, each intermediary accepts the entire message, scrutinizes the address, and then forwards the message to the next party, which can be another intermediary or the destination node.
What's especially notable about message-switching networks, and indeed happens to be one of their defining features, is that the intermediaries aren't required to forward messages immediately. Instead, they can hold messages before sending them on to their next destination. This is one of the advantages of message switching. Because the intermediate stations can wait for an opportunity to transmit, the network can avoid, or at least reduce, heavy traffic periods, and it has some control over the efficient use of communication lines.

Packet Switching

Packet switching, although it is also involved in routing data within and between LANs such as Ethernet and Token Ring, is also the backbone of WAN routing. It's not the highway on which the data packets travel, but it is the dispatching system and to some extent the cargo containers that carry the data from place to place. In a sense, packet switching is the Federal Express or United Parcel Service of a WAN.
In packet switching, all transmissions are broken into units called packets, each of which contains addressing information that identifies both the source and destination nodes. These packets are then routed through various intermediaries, known as Packet Switching Exchanges (PSEs), until they reach their destination. At each stop along the way, the intermediary inspects the packet's destination address, consults a routing table, and forwards the packet at the highest possible speed to the next link in the chain leading to the recipient.
As they travel from link to link, packets are often carried on what are known as virtual circuits—temporary allocations of bandwidth over which the sending and receiving stations communicate after agreeing on certain "ground rules," including packet size, flow control, and error control. Thus, unlike circuit switching, packet switching typically does not tie up a line indefinitely for the benefit of sender and receiver. Transmissions require only the bandwidth needed for forwarding any given packet, and because packet switching is also based on multiplexing messages, many transmissions can be interleaved on the same networking medium at the same time.

Connectionless and Connection-Oriented Services

So packet-switched networks transfer data over variable routes in little bundles called packets. But how do these networks actually make the connection between the sender and the recipient? The sender can't just assume that a transmitted packet will eventually find its way to the correct destination. There has to be some kind of connection—some kind of link between the sender and the recipient. That link can be based on either connectionless or connection-oriented services, depending on the type of packet-switching network involved.

·         In a (so to speak) connectionless "connection," an actual communications link isn't established between sender and recipient before packets can be transmitted. Each transmitted packet is considered an independent unit, unrelated to any other. As a result, the packets making up a complete message can be routed over different paths to reach their destination.

In a connection-oriented service, the communications link is made before any packets are transmitted. Because the link is established before transmission begins, the packets comprising a message all follow the same route to their destination. In establishing the link between sender and recipient, a connection-oriented service can make use of either switched virtual circuits (SVCs) or permanent virtual circuits (PVCs):

o        Using a switched virtual circuit is comparable to calling someone on the telephone. The caller connects to the called computer, they exchange information, and then they terminate the connection.