Infoblast Home

Media History and Overview

Ethernet originated in the 70's from efforts at Xerox's PARC to create a new method for connecting computers together for sharing programs, data and information

The first version of Ethernet was soon replaced by Version II that was developed by a joint effort between Digital Equipment Corp, Intel Corp and Xerox Corp. This is also know as the DIX Ethernet standard














In this example both PC A & PC B believed that the network was available and started to transmit nearly simultaneously. Neither PC heard the others transmissions right away due to the propagation delay of the bits traveling on the cable. This causes a collision. In this case PC B will detect the collision. PC B stops transmitting and sends a JAM signal telling all devices on the cable to back off and try again. PC A hears the JAM signal and stops transmitting. Each device waits a period of time based on the standardized back off algorithm and tries again.

The minimum packet length on an ethernet network is 512 bits (64 bytes). This minimum length ensures that a stations is still transmitting a frame when all other stations on the ring have had an opportunity to see the frame and raise a collision alert if necessary. Collisions raised by other stations tell a transmitting station that its frame was not delivered and it will have to try again.


In this example PC A has just finished transmitting a 64 byte frame onto the cable. PC B did not detect PC A's frame even though it met the minimum size requirement. This can be caused by an out of spec cable plant. Normally a 60 byte frame would have filled up the entire cable and PC A would have still been transmitting when PC B attempted to transmit. A collision would have occurred and PC A would have known that it would have to retransmit its frame.

In this case PC A has finished transmitting the frame and removed it from its hardware buffers before PC B started transmitting its own frame. Eventually PC B will detect PC A's frame and a collision will result. PC A will detect the collision signal but will not know that it was referring to the frame it just sent. PC A will assume another station was the cause of the collision. As far as PC B is concerned the frame it detects is just another normal collision. PC B has no knowledge that this was in fact a frame that PC A thought it had transmitted successfully.



In this example both PC A has transmitted the first 60 bytes of its frame on the cable. As PC A has not heard a collision for this frame yet it will continue sending the balance of the frame. However PC B has not yet heard this frame. Possibly due to an out-of -spec cabling system. PC B begins transmitting its own frame. PC B will soon discover its mistake when it hears the frame from PC A on the cable. PC B immediately issues a JAM signal. PC A will hear the JAM and stop transmitting. PC A will log a Late Collision because it should not have been able to get 60 bytes of a frame onto the cable before hearing a collision.

In fact the time it takes to transmit a 64 byte frame onto a cable has to be greater than propagation delay of sending a bit from from one end of the cable plant to the other and back. This is because a station at one end may start sending a frame while a station at the other end may also start sending a frame at the same time. Each station must detect the other stations' frame and send a JAM signal that must reach the originating station while it is still transmitting its frame.

In other words a cable plant can only be <32bits long in order to work successfully.







The station sends frames to itself without inserting itself into the ring. The frames are looped at the MAU. If the test passes the station sends a 5 volt signal down the lobe cable. This opens the relay and inserts the station onto the ring.

If no active monitor is present this station will become the active monitor.

Duplicate MAC addresses are not allowed on the same ring. If the station detects another station with the same address already on the ring it will not insert itself into the ring.

This is the process where each station on the ring discovers its upstream and downstream neighbors. This information is used for isolating fault domains.



Copyright 1992-1999 © Cisco Systems Inc.