Supervisor: Dr. Steve Vickers
Student: Lim Ming Hai
Course: Information Systems Engineering II,
Imperial College
Coursework: Article 1 of SURPRISE '97
e-mail address: mhl@ic.ac.uk
Date of submission: 20th May 1997
Abstract
An increasing number of computers nowadays are being connected up into networks and Ethernet is the most popular way of doing so. With Ethernet, fair access to the shared channel is ensured by Carrier Sense Multiple Access with Collision Detection (CSMA/CD) rules. Data is delivered between stations using Ethernet frames which are divided into fields containing destination address, source address and data. Thousands of computers can be connected together using hubs. The original standard for Ethernet operated at 10-Mbs but increasingly, there is a move towards 100-Mbs and 1000-Mbs Ethernet technology.
Origins
The most popular Local Area Network (LAN) technology in the world today is Ethernet. With an estimated 70 million nodes installed world-wide, its popularity is an order of magnitude greater than its closest rival, token ring. Some of the reasons for its prevalence include the easy availability of the specifications and rights to build Ethernet technology, the ease of use and robustness of the system. Also, Ethernet provides support for a wide variety of computers.
The origin of the Ethernet can be traced to the Xerox Palo Alto Research Center where it was invented in the 1970s by Dr. Robert M. Metcalfe. The very first Ethernet system (called experimental Ethernet) could only manage 3-Mbs and was initially designed to support research being done at Xerox into futuristic offices.
In 1980, Digital Equipment Corp., Intel and Xerox (DIX) formed a consortium which promoted formal specifications for Ethernet. Experimental Ethernet was improved into an open, production-quality Ethernet system that operated at 10-Mbs and submitted to the Institute of Electrical and Electronics Engineers (IEEE) for standardization.
In 1985, the IEEE standard was published as "IEEE 802.3 Carrier Sense Multiple Access with Collision Detection (CSMA/CD) Access Method and Physical Layer Specifications". Since then, all Ethernet equipment has been built according to this world-wide networking standard.
Operations
A complete working Ethernet system can be broken down into three components:-
In an Ethernet system, data signals from a computer (or station) are transmitted serially down a shared channel to every station of the system. Since each station acts independently of all other stations, Carrier Sense Multiple Access with Collision Detection (CSMA/CD) is used to control access to the shared channel.
Carrier Sense refers to the detection of a signal (or carrier) on the channel. When an interface is transmitting, all other interfaces must wait till the carrier has completed before attempting their own transmission. Multiple Access means that each interface in the system has an equal priority to send a signal. Collision Detection occurs when two or more interfaces start transmitting at the same time. The result is that all transmissions are stopped and the interfaces have to reschedule their transmission. An algorithm is used to choose a random time interval before frames are resent. The primary aim of the CSMA/CD protocol is to ensure that all stations have fair access to the shared channel. However, there is no guarantee of reliable data delivery. Electrical noise in cables may corrupt frames and we need to use high-level network protocols to establish acknowledgement mechanisms.
The Ethernet frame is the packet that delivers data between stations. Each frame is made up of bits which are organized into different fields. The four main fields are the destination Ethernet address field, source Ethernet address field, data field and frame check sequence field. The address fields are 48 bit long. In the case of a broadcast address, the destination address is made up of all "1"s and all interfaces will read in the frame. The source address is normally pre-assigned to each Ethernet interface when it is manufactured. The data field has a variable size from 46 to 1500 bytes and short packets must be padded up to at least 46 bytes. The frame check sequence field is a 32 bit Cyclical Redundancy Check (CRC) which checks the integrity of the bits in the frame to ensure that the frame has arrived intact.
The TCP/IP protocol so widely used on the internet is an example of a high-level protocol software that can be used by Ethernet to send application data from one station to another. The protocol packets are contained in the data fields of the Ethernet frames. An Ethernet system can carry many different varieties of protocol data (e.g. AppleTalk), in fact it makes no difference to Ethernet what it carries.
A repeater is a signal and retiming device that can be used to link large numbers of Ethernet segments to form a large Ethernet LAN. However, an Ethernet system will not function correctly if there are loop paths present so it must be guaranteed that segments are never connected in a loop.
The figure above demonstrates how repeaters can be used to connect stations and segments together. When a signal is sent from any station, it travels over that station's segment and is repeated to all other segments, thus ensuring that it is transmitted to all stations on the Ethernet channel.
Round trip time refers to the amount of time needed for a signal to get from one end of a media system and back. The maximum round trip time of signals on the shared Ethernet channel has to be strictly limited to ensure that each interface can detect signals on the channel in the time specified by the MAC system. This is the main limiting factor to how long a network segment can be.
To allow Ethernet to support thousands of stations while still meeting the round trip time, a packet switching hub is used to link individual Ethernet LANs together. This is possible because the round trip timing rules for each Ethernet LAN stops at the switching hub ports.
Fast Ethernet
To meet growing demands for speed, there are two LAN standards that can transmit Ethernet frames at 100-Mbs. 100BASE-T Fast Ethernet involves speeding up the original Ethernet system to 100-Mbs while retaining the CSMA/CD MAC mechanism. 100VG-AnyLAN requires the use of a new MAC mechanism which allows the use of demand priority. The latter can also be used to transport token ring frames.
Fast ethernet packets travel at a speed 10 times greater than Ethernet resulting in a corresponding reduction in bit-time (the time needed to transmit a bit on the Ethernet channel). Other fundamental aspects of Ethernet (e.g. frame format, MAC rules, amount of data in frame) are unchanged and together with the Auto-Negotiation mechanism of media speed allow vendors to supply dual-speed Ethernet interfaces which can automatically switch from 10-Mbs to 100-Mbs.
To connect a station to a Fast Ethernet network, the computer must have an interface that will form up and send the frames. The interface is then attached to the media system using an outboard Medium Independent Interface (MII) cable, Physical Layer Device (PHY) and Medium Dependent Interface (MDI). The MII makes the signalling differences among different media segments transparent to the Ethernet chips. Fast Ethernet systems have a PHY and MDI specifically designed for each media type.
The next step for Ethernet will be Gigabit Ethernet, which as the name indicates involves moving packets 10 times faster than Fast Ethernet.
References
Charles Spurgeon
Ethernet Configuration Guidelines
Peer-to-Peer Communications, Inc.
ISBN: 1-57398-012-9
John E. McNamara
Local Area Networks, An introduction to the technology
Digital Press, 1985
ISBN 0-932376-79-7
Digital Equipment Corporation
Network Troubleshooting Guide
Digital Press, August 1990
http://wwwhost.ots.utexas.edu/ethernet/ethernet-home.html
Welcome to Charles Spurgeon's Ethernet Page
http://www.networking.ibm.com/atm/atm25fe.html
Desktop ATM versus Fast Ethernet
A White Paper
http://cesdis.gsfc.nasa.gov/linux/misc/100mbs.html
100Mbs Ethernet Technology Page
http://alumni.caltech.edu/~dank/fe/
Dan Kegel's Fast Ethernet Page (15 Apr 97)
http://www.iol.unh.edu/consortiums/fe/index.html
Fast Ethernet Consortium
http://cesdis.gsfc.nasa.gov/linux/misc/NWay.html
An Introduction to Auto-Negotiation
Bill Bunch, February 1995
http://members.aol.com/dhawley/isoenet.html
Dave Hawley's isoEthernet Page