networking curriculum
support CD for senior
secondary studies

networking > osi model > basic problems

Basic network problems

On a network segment with multiple nodes, only one message or packet can pass along a network cable at a time, so while one node is transmitting, every other node (workstation, server, printer etc.) must wait. If a network had hundreds of nodes, you can imagine the delays that would happen as each computer fought for a chance to 'speak'. So a network needs a way to manage network traffic.

Also, when a packet is sent across the network segment, the packet is sent to every node connected to the network cable to let each node check to see if the packet is addressed to it. This means that workstations constantly have to receive and check packets that are irrelevant to them. If the number of packets flooding a workstation gets too high, this can lead to network congestion.

It's like every person in a shopping centre using the public address system to talk to each other. Not only does every other shopper have to listen to each message in case it's meant for them, but no one else is able to speak while the PA is in use.

So - how do networks handle these problems?

The problem of controlling which node is allowed to transmit is handled by the network technology being used (e.g. Ethernet or Token Ring).

Network congestion is handled with clever network architecture - the way the network is designed. Congestion is avoided by using segments and switches.

There are various "shapes" a network can take. These shapes are called topologies and are discussed in more detail in another section, but here's a brief starter.

The simplest way to connect a group of computers is to connect them all to the same cable like a chain of Christmas tree lights. This is called a bus topology.

You will notice that the packet visits every node in turn so the node can check if the packet was addressed to him. This means that it is easy to overwhelm the computers with irrelevant packets, causing congestion. It's also important to note that if there were a break in any part of the cable, all of the computers connected to that cable would stop receiving packets - just like a string of Christmas lights all failing because one bulb blows.

Another way of connecting computers together is like this, called a star topology

With a star topology a single incoming cable connects to a device (either a hub or a switch) which sends a copy of the packet down multiple outgoing cables, which lead to PCs. In this way, if a cable leading to a single PC failed, only that PC would be out of action; the other PCs would be unaffected. Of course, a break in the incoming cable would cause all of the PCs to lose their network connection. The group of PCs connected to the hub is called a network segment. Once again, however, we still have the problem of each PC having to receive and check packets that are irrelevant to them.

Hubs are dumb creatures: they pass every network packet they receive to every node in their segment, regardless of whether the packet is addressed to one of their nodes or not. On a large network, the number of irrelevant packets received by each node is enormous, and delays caused by congestion can rapidly cripple a large network.

Switches ('switching hubs') have now largely replaced hubs. Switches have the intelligence to examine each packet they receive from the network, and only pass on packets that are addressed to one of the nodes in their segment. Switches greatly reduce network congestion.

The red line in the image above shows the path of a packet being routed directly from a server to a single node through switches. Nodes on blue lines do not see the packet at all. Switches can dramatically reduce the amount of irrelevant network traffic a node sees. Most of the time, nodes only see packets that are intended for them alone. (Occasionally, a server may need to deliberately send a 'broadcast' message to every workstation.)

Hubs and switches come in a variety of sizes (from 4 to 50 ports), speeds (10, 100 or 1000 megabits per second), management modes (managed switches can be controlled remotely), and can be stackable so that two or more of them can be linked to behave like a single, very large switch or hub.

Bridges were devices used to connect network segments. They only passed on packets relevant to their own segment. Nowadays, switches are used as bridges.

By using switches, and dividing large networks into separate segments, network congestion can be effectively managed.

This still leaves the problem of deciding which node gets to speak at a particular time. This is examined in the next section.

:: Home |  History of Networking |  Why set up a Network? |  OSI Model |  Connecting up |  Protocols ::
:: Technology |  Types of Networks |  Topologies |  Issues |  Design Factors |  Economic effects ::
:: copyright |  credits |  sitemap ::

© 2003 Victorian Information Technology Teachers Association
  h t t p : / / w w w . v i t t a . o r g . a u /