Chapter 6

Ethernet Fundamentals

These notes guide you through the understanding of Ethernet Frames.

Introduction

Ethernet Overview
Ethernet Frames
        Ethernet II
        Ethernet 802.3

Ethernet Errors

Summary

Introduction

On completion of these notes you should...

• Understand the basics of Ethernet technology
• Describe the structure of Ethernet frames
• Understand the different types of frame errors that can occur on Ethernet

Ethernet Overview

Ethernet refers to a family of protocols specified under IEEE 802.3 specifications.

The original version of Ethernet was designed in 1976 by Dr. Robert M. Metcalfe and his co-workers at Xerox. Originally, Ethernet transmitted at a data rate of 10Mbps over thicknet coaxial cable. In the 1980's the IEEE published standards to ensure that Ethernet complied with the OSI model, modifying the original Ethernet a little. These standards are referred to as the 802 standards.

Currently, we have the following Ethernet data rates, carried over optical fiber and twisted-pair cables:

IEEE 802.3		10 Mbps 10Base-T Legacy Ethernet
IEEE 802.3u		100 Mbps Fast Ethernet
IEEE 802.3z		1000 Mbps Gigabit Ethernet
IEEE 802.3ae		10 Gbps Gigabit Ethernet

All the Ethernet standards are essentially compatible with each other. An Ethernet frame should be able to travel from Legacy Ethernet to Gigabit Ethernet without any problems. This flexibility of Ethernet is why it has become so common. As data rates increase and new Ethernet standards emerge for these faster rates, the older Ethernet standards can still be used. Whenever newer faster Ethernet technologies develop, the IEEE just add as new 802.3 specification.

Ethernet Frames

Information transmitted onto the wire on Ethernet LANs are encapsulation into Layer 2 units called frames. There are four choices of Ethernet frame type. You can have Ethernet II, Ethernet 802.2, Ethernet 802.3 and Ethernet SNAP.

Ethernet II is normally used for TCP/IP whereas Ethernet 802.2 and, Ethernet 802.3 is the most common frame types used on a NetWare network. NetWare servers before version 3.12 default to Ethernet 802.3. The newer versions default to Ethernet 802.2. Ethernet SNAP is used for AppleTalk.

All Ethernet frames consists of a standardized set of bits used to carry data over the system. In other words they are structured in particular ways.

Start information - all frames contain a starting sequence of bytes that signal a frame is on the way.

Naming information - all frames contain naming information such as the name of the source node (MAC address) and the name of the destination node (MAC address).

Length/Type Field - some technologies include a length field specifying the exact length of a frame in bytes. Some frames have a type field instead, specifying the Layer 3 protocol making the sending request.

Data Field - all frames contain data

FCS Information - frames usually contain a number which allows the receiving device to check if the frame is error free. This is called the Frame Check Sequence number. The most common method of generating this number is by using the Cyclic Redundancy Check (CRC). The sending device calculates the CRC number and includes it in the frame. The receiving end recalculates the CRC number and compares it with the number in the frame. If the number does not match, it requests retransmission of the frame.

Ethernet II Frame

The first version of Ethernet, also known as Ethernet I was developed by Xerox. Ethernet II is a revised version of Ethernet rewritten by with Digital Equipment Corp, Intel and Xerox. Ethernet II is also known as DIX.

An Ethernet II frame contains the following fields.

• Preamble: - 8 bytes, an alternating pattern of ones and zeros that tells receiving station that a frame is coming and provides a means to synchronize the receiver with the incoming bit stream.
  

• Destination: - 6 bytes, an address field identifying the destination host
  

• Source: - 6 bytes, an address field identifying the sending host
  

• Type: - 2 bytes, a field indicating the type of frame.
  

• Data: - 46 to 1500 bytes, a field containing the data stream.

• Frame check sequence (FCS): - 4 bytes, a sequence containing a 32-bit cyclic redundancy check value (CRC), which is created by the sender and is recalculated by the receiver to check for damaged frames.

Ethernet 802.3 Frame

In the 1980's the IEEE standardized Ethernet, developing a new 802.3 frame type. This frame type is most commonly used on NetWare networks.

The structure of an 802.3 frame type is shown below.

• Preamble: - 8 bytes, an alternating pattern of ones and zeros that tells receiving station that a frame is coming and provides a means to synchronize the receiver with the incoming bit stream.
  

• Start of frame delimiter: - 1 byte, an alternating pattern of ones and zeros, ending with two consecutive 1-bits.

• Destination: - 6 bytes, an address field identifying the destination host
  

• Source: - 6 bytes, an address field identifying the sending host

• Length/Type: - 2 bytes, a field indicating either the length of frame or type of frame.

• Data: - 46 to 1500 bytes, a field containing the data stream.

• Frame check sequence (FCS): - 4 bytes, a sequence containing a 32-bit cyclic redundancy check value (CRC), which is created by the sender and is recalculated by the receiver to check for damaged frames.

In the 802.3 version of Ethernet, an extra Start Frame Delimiter (SFD) was added after the preamble.

it is important to realize that the 802.3 Length/Type field supports two different uses. To determine if the field is a length or a type, the receiving node must examine value of the number in the field. If the value is less than 0x600 (1536 decimal) it is interpreted as length. If the value is equal to or greater than 0x600 then the frame is interpreted according to the Ethernet II type code indicated. The type code specifies the upper-layer protocol to receive the data.

~~Activity~~

Ethernet Errors

This section briefly discusses the main error types that can occur on Ethernet LAN's.

~~ Long Frame ~~

A long frame is a frame where the size is over the legal maximum size of 1518 octets. Long frames are sometimes confused with Jabber.

~~ Jabber ~~

Jabber is defined in the 802.3 standard as an excessively large transmission lasting from 20,000 to 50,000 bit times.

Additionally, Jabber is often reported by diagnostic reporting software when a frame is greater than the maximum legal size of 1518 bytes, which may be much smaller than the 802.3 definition of a frame lasting 20,000 to 50,000 bit times.

A node on a network that is continuously transmitting, or transmitting improperly for short bursts, probably due to a bad transmitter on the NIC, is sometimes referred to as a jabbering node.

~~ Short Frame ~~

A short frame is a frame where the size is under the legal minimum size of 64 octets but the FCS checksum is valid.

~~ Runt ~~

A runt is a slang term sometimes used to refer to short frames and sometimes used to refer to collision fragments.

~~ FCS Errors ~~

Before a station sends a frame, it uses an algorithm to calculate a Cyclical Redundancy Check (CRC) value. The value that pops out from the algorithm depends on the length of data in the frame. This value is inserted into the Frame Check Sum (FCS) field of the frame.

The receiving station recalculates the CRC value and compares this value with the value in the FCS field of the frame. If the frame is altered between the source and destination, the CRC values will not match. The frame will then be discarded.

~~ Alignment Errors ~~

An alignment error is caused when an incoming packet does not end on an octet boundary. This means that extra bits(less than 8) that do not form an octets will be left over. These extra bits are discarded from the frame and since it is unlikely for the FCS checksum to match, an alignment error will be reported.

~~ Range Error ~~

A range error is reported when a frame arrives with a value in the Length field that does not match the actual number of octets in the data field.

The length of data in a frame must be between 46 bytes to 1500 octets. So this error also occurs when the value in the Length field indicates the unpadded data size is under the legal minimum size. It also occurs when the value in the Length field indicates the data size is over the legal maximum size.

~~ Ghost ~~

Ghosts are peculiar manifestations and are generally accepted to mean noise detected on the cable that appears to be a frame, of at least 72 octets in length, but lacking a valid SFD.

Wiring problems, such as ground loops are usually the cause of ghosts. With some types of signal noise, nodes and repeaters may be fooled into thinking a valid frame is arriving. However, a frame that is valid never actually arrives and so is eventually discarded by a node. Repeaters usually propagate this ghost noise onto other segments.

Summary

On completing these notes you should have learned the following key points:-

• The basics of Ethernet
• Ethernet naming rules
• The structure of Ethernet II and 802.3 frames
• Ethernet frame field names and purposes
• The types of frame error that can occur
• The difference between long frames and jabber
• The difference between short frames and runts