Figure 3.1 A shared medium Ethernet network
Introduction
History
Ethernet -
Shared and Switched
Medium Access Method
Reception of Frames
Frame formats
Conclusion
Resources
References
This lecture will examine:
Ethernet was developed by Xerox in mid 1970s and was the basis for IEEE 802.3.
DEC, Intel and Xerox released Ethernet version 2.0 which was compatible with IEEE 802.3.
Ethernet refers to any CSMA/ CD LAN, including IEEE 802.3.
Ethernet - Shared and Switched
Shared
The earliest Ethernet networks were implemented using coaxial cabling (10 Base 5) that ran from one end of the network to the other. Stations could be added at any point in the cable using a vampire tap, a tool that clamped to the coaxial cabling and pierced the outer layers of the coax to make a connection with the central wire strand. Shared Ethernet is good for networks carrying sporadic communications and occasional heavy traffic.
Figure 3.1 A shared medium
Ethernet network
In figure 3.1, it can be seen that the computers are connected to one cable, this is the network. The coaxial cable has just one central conductor so that it can carry only one message from any one computer at any one instant in time.
Although the network may be rated as 10 Mbit/ second, this speed would only ever be achieved if there was only one computer connected to the network. By allowing many computers to be attached to the network, the bandwidth of 10 Mbit/ second is shared between the nodes (computers) that form the network.
This means that each computer is able to have all of the bandwidth, some of the time. Clearly, if we have more than one computer connected to the same piece of cable, there would be chaos if the computers were allowed to broadcast a message whenever they wanted to. This would lead to a situation where messages from two or more different computers were transmitted onto the network cabling simultaneously and these messages would mix together forming an unintelligible signal that could not be understood by the intended recipient(s).
A set of rules was devised and implemented in the computers to allow the nodes to share the cabling with each other.
The choices of sharing method were constrained by the following:
1. It is not possible to predict when a computer will be ready to transmit a message.
2. It was not desirable to have one central computer to control the access of the other nodes to the cabling. The time required to poll each node in turn to see if it had anything to transmit would be detrimental to the network bandwidth. Also, if the central computer was to fail, the entire network would fail.
Given these two conditions, Xerox worked out a set of rules that could be implemented into every machine in the network. This meant that each computer would act on its own and if a node failed it would not affect the rest of the network.
This protocol is known as CSMA/ CD, Carrier Sense Multiple Access with Collision Detection.
Switched
Today it is unlikely that you will find a shared medium Ethernet network anywhere but in a legacy installation (i.e. one that was installed a long time ago). The popularity of the Ethernet LAN led to massive take-up by business and added to this the ease of joining more nodes to a network led to worse and worse performance as numbers of nodes attached to the network backbone increased. Each station would have all of the bandwidth for even less of the time.
The nature of the CSMA/ CD protocol meant that as the number of nodes increased, so the probability of collisions increased; this in turn reduced the throughput of data as the protocol dealt with the collision(s). An attempt to alleviate the problem was made using a computer configured as a bridge. The bridge separated different sections (known as segments) of the network and filtered traffic on the backbone, keeping network traffic to the segments that it needed to cross only. This reduction in traffic improved bandwidth temporarily, but was only an intermediate solution.
Advances in cabling and hardware have led to the introduction of Ethernet networks connected using Cat 3 and Cat 5 cabling, both of which are twisted pair, rather than coaxial cabling. Devices known as switches have also been introduced and their cost is dropping in real terms. What the switch manages to do is to collapse the backbone of the network (the coaxial cable) into a box with several sockets on the outside for nodes to be plugged into.
All recent Ethernet implementations employ switches and Cat 5 twisted pair. This has changed the topology of the network from bus into star.
Fig 3.2 Nodes on Ethernet
network form a star topology by connecting to a switch
Strictly speaking, when using a switch the MAC method
is no longer needed because collisions cannot happen but is still used
to retain compatibility with older sections of the network that may still
be sharing a backbone. the operation and technology of the switch will
be dealt with later in this course.
CSMA/ CD operates as follows:
In this case, two stations that began to transmit while the line appeared to be clear will, after a short time, realise that another station was also transmitting. Each station will realise this when the other station's message propagates along the shared medium and reaches the opposite station.
Both messages are damaged and the stations stop transmitting and will transmit at some later time. This time is determined by a backoff algorithm.
The transmissions of nodes in an Ethernet network are placed in structures known as frames. a frame is just a logical structure for containing transmitted data and control information to assist in delivery of the frame.
In an Ethernet/ IEEE 802.3 LAN, all stations are joined to the same cable and will receive all frames from all other nodes. Frames are examined to determine if the message is intended for that particular station. If so, the frame is passed to a higher protocol layer for processing, else it is ignored.
Ethernet and IEEE 802.3 are implemented in hardware, generally in the form of an interface card that slots into an expansion socket on the motherboard.
Each IEEE 802.3 specification has a name that specifies its characteristics.
Fig 3.3 Depicting how the
naming convention of IEEE 802.3 operates
| Ethernet |
|
|||||
| 10Base5 | 10Base2 | 1Base5 | 10BaseT | 10Broad36 | ||
| Data Rate Mbps | 10 | 10 | 10 | 1 | 10 | 10 |
| Signaling method | Baseband | Baseband | Baseband | Baseband | Baseband | Broadband |
| Max segment length, m | 500 | 500 | 185 | 250 | 100 UTP | 1800 |
| Media | 50 W coax thick | 50 W coax thick | 50 W coax thin | Unshielded twisted pair UTP | Unshielded twisted pair UTP | 75 W coax |
| Topology | Bus | Bus | Bus | Star | Star | Bus |
Table 3.1 Ethernet v 2.0 and IEEE 802.3 Physical characteristics.
From the diagram below it can be seen that the format
of an Ethernet frame is almost identical to that of an IEEE 802.3 frame.
The only difference is the use of the 2 byte field before the data. Ethernet
uses this as a Type field whereas IEEE 802.3 uses this field to denote
the length of the data field, which is variable according to the amount
of data to be sent.
Figure 3.4 The format of Ethernet
and IEEE 802.3 frames. The numbers above each frame are the bytes allocated
to each section
It can be seen from fig 3.4 that a frame consists of:
Shared Ethernet needed a set of rules to avoid collisions of signals or frames on the network. This set of rules is known as CSMA/ CD.
Switched Ethernet is the standard today, does not allow for collisions.
CSMA/ CD is not strictly needed in switched Ethernet, but still implemented for backward compatibility.
There is very little difference between Ethernet and IEEE 802.3 and their respective frames have little difference either.
Ethernet for Industrial Control - An Ethernet White Paper - April 21, 1998
PC Support Handbook, D Dick, Dumbreck Publishing 1998, pp 433 - 434
Internetworking Technology Overview, Cisco Systems, Chapter 5
Back to top of page
© MM Clements 2001