Lecture 3 (Network Types)

Slide 1 [Introduction]
Hopefully by now everyone should have registered for the module, and should have received a confirmation email message. Updates on the module will be sent by email. Everyone who is studying the module is treated the same, and you should make sure that you attend the practical session for one hour every week. In most weeks, you will complete a worksheet, which will possibly be followed by a small on-line test based on a previous unit. In the coming week you will complete Worksheet 2, and take a small on-line test on Unit 1.

By now you should understand the operation of the module. Each of you should have a one hour practical session booked for you. If you do not have this make sure you attend one of the practical which run from Monday to Thursday. Also if you have any questions on any of the material please send an email question, and we'll try and answer it.

Today we will look at some of the fundamental principles of networks, especially in relation to network topology as this gives us a key in understanding how the network will perform, especially under extreme conditions. It must be remembered that there are many different types of networks, each with a specific function, and each performing to a given criteria. For example networks that contain control equipment in an industrial plant or which are in control of safety critical systems have different requirements to ones which connect a home computer to the Internet, or even one that connects users to their email system. This is because certain types of data in a safety critical network (such as alarm events) must have priority over other types of data (such as email traffic). Thus there is no definite networking topology that will fit all applications.

Slide 2 [Interconnection of networks]
Networks have grown extremely quickly within organisations, and we need to have methods which allows us to isolate one network from another. This is required for many reasons, such as:

	Security. The data and traffic on certain networks requires to be kept secret, thus there must be some way of stopping certain external devices from getting access to it. In most case a firewall is typically used for this purpose.
	Concurrency. This allows different networks to communicate at the same without affecting other networks.
	Improved organisation. Most organisations have a certain structure, which should typically be mirrored in the layout of the network structure.

There are two main devices which are used to segment a network into isolated network segments, these are routers and bridges. They have different functions, and operate a different layers of the OSI model. First let's look at the topology of networks.

Slide 3 [Network Topologies]
A topology defines who the nodes on a network actually connect to each other. The three topologies are: ring, star and bus. There are three main networking topologies which can be mixed together to produce a hybrid topology.

Slide 4 [Tree topology]
A network can also be created in a tree structure, where nodes connect to network hubs. Not that the hub itself is like a star connection, as it can be viewed as a central point of connection. Inside the hub, the network topology could either be star, bus or ring-based. For example an Ethernet hub is a single point of connection for networked nodes. If it were to fail then all the nodes which connect to the hub will not be able to communicate. Inside the hub the actual network created is a bus type.

Hubs are good devices to have on a network, especially if users who normally share network traffic connect to the same hub, as the communications can stay local within the hub. As much as possible users who work together should be congregated around workgroup hubs.

Slide 5 [Star network]
With a star topology there is a central server which routes data between the other nodes on the network. This has the advantage that the data transfer between the nodes and the server will be relatively small, as it only carries the traffic between an individual nodes and the server. Also a fault on one of the nodes will not bring-down the whole of the network. Unfortunately, a star network, though, is extremely dependent on the central server. If it were to develop a fault or slow-down, if will affect the whole of the network.

Slide 6 [Bus network]
In a bus network all the networked nodes connect to a common network connection. Thus the nodes must contend to get access to the network. There can thus be no guarantee that a node can ever get access to the network, whenever it needs to, as it may have to wait for another node to complete its transfer.

The great advantage of a bus network is that it is extremely easy to add and delete networked nodes to and from the network, as all that is required is that they connect to the common bus.

Slide 7 [Token passing ring network]

As we have seen a star network suffers from a major focus on the network server, and the bus network suffers in that nodes must contend to get access to the network. Ring networks are one of the best techniques, and are typically used where nodes must be given an equal share of the available data traffic.

Ring networks require two connections for each node (up-stream and down-stream), and join together to make a ring. On a single ring, data is only transferred in the one direction (such as clockwise or counter-clockwise). The up-stream is defined as the output to another node, and the down-stream is the input from another node.

As a node must release the token when it has completed it transfer, all the nodes will have an equal chance to get access to the network. This could cause a slight problem if one node want to transmit large amounts of data, and no other node has data to send. This is because the node that wants to transmit data must release the token after it has transmitted its data frame. It will then have to wait for the nodes to transmit the token around the ring, until it comes back to the original node, again. This problem can be overcome by creating a timed token, where a node can keep the token for a certain time limit, before it must release it. This could be expanded on, so that certain nodes can have a longer timed token than others.

Slide 8 [Token Ring]
Token passing is rather like attending a busy conference, where someone in the audience cannot speak until they get a token, which is passed from person to person. Once they get it they will be allowed to speak. In a ring approach the token would ripple through the audience in a fixed way.

In token passing, the token circulates around the ring, until a node captures it. It can then transmit its data frame. This data frame is passed from node to node, around the ring. Once the destination node reads the data frame it sets one of the bits in the data frame to identify that it has read the data in it. This provides an acknowledgement for the sender. The data frame then returns back to the sender, who then releases the token back onto the network.

The great advantage of token ring networks is that is allows all the nodes an equal chance to get access to the network, at a given time. Unfortunately it has suffered from two main problems:

FDDI also has a method of loop back which allows nodes to detect that there is a break in both the rings, and create a ring with the remains of the two rings.

Slide 9 [CSMA/CD]
In a bus network, nodes must contend to get onto the network, as only one node can communicate at a time. Thus there must be some mechanism to stop more than one node communicating at a time. This is achieved with CSMA/CD (Carrier Sense Multiple Access/ Collision Detection). The carrier sense part allows nodes to sense if there are no other nodes communicating at a given time, and the collision detection allows nodes to detect if more than one node is communicating at a time.

In CSMA/CD if two nodes try to communicate at the same time, they detect this, and both of them transmit a jamming signal to the whole of the network. No other nodes can then transmit while the jamming signal is being transmitted. Each of the nodes who caused the collision then wait for a random time, and one of them will get access to the network before the other. This CSMA/CD is thus a contentious network.

Slide 10 [Routers, bridges and repeaters]

An important concept on networks is the differentiation between a network address, and a physical address. The physical address is typically randomly assigned, and is hard-wired into the network adaptor. It would be impossible to know someone's destination physical address, thus a network address is used to give some information on where the node is actually located. On the Internet the network address is known as an IP address, and on an Ethernet network the physical address is known as the MAC (Media Access Control) address.

Routers operate on the network address, and bridges operate on the physical address. Both devices can detect if the transmitted data is destined for a network outwith the current network.

Slide 11 [Routers, bridges and repeaters]

Slide 13 [Cabling]

Slide 14 [Maximum for twisted-pair]