Ethernet Background and Encapsulation Overview

The motivation behind the development of a computer network came from Xerox wanting to interconnect some of the first personal computers at its Palo Alto Research Center (PARC) and the world's first laser printer so that all of the PARC's computers could print with the same printer. Prior to this, the number of computers at any single facility never exceeded two or three. This was the first instance in which hundreds of computers in the same building required intercommunication.

Ethernet solved two of Xerox's challenges:

• The network had to connect hundreds of computers within the same building.

• It had to be fast enough to support the fast, new laser printer.

Later on, because of the combined efforts of Digital Equipment, Intel, and Xerox, Ethernet became a standard. Ethernet is now the world's most widely used LAN protocol.

Ethernet uses carrier sense multiple access collision detect (CSMA-CD). The different parts of this protocol are as follows:

• Carrier sense Before transmitting data, stations check whether other stations are already transmitting over the multiaccess wire. If other stations are not transmitting, the station can transmit data or wait.

• Multiple access All stations are connected to a single physical wire or a single data path.

• Collision detect If a collision is detected because two stations transmitted data into the wire simultaneously, both stations stop transmitting, back off, and try again later after a random delay.

The base of Ethernet technology is the Ethernet frame. An IP datagram, for instance, is encapsulated and transmitted in a standard Ethernet (Type II) frame. The frame header is 14 bytes long6 bytes of destination address + 6 bytes of source address + 2 bytes of frame typefollowed by the data portion and completed by 4 bytes of the frame check sequence (FCS). Figure 4-1 shows the fields of the original Ethernet Type II (Ethernet II) frame format.

Figure 4-1. Ethernet Type II Frame

[View full size image]

The following explains the fields in an Ethernet II frame:

• Destination Address The destination address can be a broadcast address 0xFFFFFFFFFFFF, a specific 48-bit unicast address based on the destination node's MAC address, or a multicast address. You can discover this MAC address from the source address field of a message during protocol synchronization.

• Source Address The source address is the sender's 48-bit MAC address.

• Ethernet Type The Ethernet Type field is used for higher protocol identification.

• Data The Data field contains encapsulated data (such as an IP packet). The valid length ranges for Ethernet II are between 46 and 1500 bytes.

• FCS The FCS field contains a 32-bit cyclic redundancy check (CRC) value, which checks for damaged frames.

The original Ethernet II frame format had some shortcomings. To allow collision detection, the 10-Mbps Ethernet required a minimum packet size of 64 bytes. That meant you needed to pad short frames with 0s. Thus, higher-layer protocols needed to include a Length field to discriminate the actual data from the padding. As a consequence, the original Ethernet frame was changed to include a Length field and to allow for Ethernet to interwork with other LAN media.

Fortunately, the values assigned to the Ethernet Type field (0x0600 XNS [Xerox], 0x0800 IP [Internet Protocol], and 0x6003 DECNET) were always higher than the maximum frame size with a decimal value of 1500. The 802 committee solution to the task of providing a standard that did not depend on the behavior or characteristics of higher layer protocols was 802.3. 802.3 replaced the Ethernet Type field with a 2-octet length field. The way to distinguish an Ethernet II from an 802.3 frame is by inspecting the Type/Length field:

• If the value of the field is higher than 1500 decimal, the field represents an Ethernet Type and is Type II.

• If the value of the field is lower than or equal to 1500 decimal, the field represents a length and is 802.3.

In addition, a new form of packet type field was needed, so a Logical Link Control (LLC) header with destination and source service access point (DSAP and SSAP, respectively) and control fields follow the Length field for higher-protocol identification (see Figure 4-2).

Figure 4-2. 802.3 Frame Format

[View full size image]

The new fields are as follows:

• Length This is the length of the frame excluding the preamble, FCS, addresses, and length field.

• DSAP A value of 0xAA indicates Subnetwork Access Protocol (SNAP).

• SSAP A value of 0xAA indicates SNAP.

• Control The Control field specifies the type of LLC frame.

Figure 4-2 also shows an IEEE 802.3 SNAP frame format that is indicated by the DSAP and SSAP values and includes the SNAP field. The SNAP header includes 3 bytes of vendor code and 2 bytes of local code. A vendor code of 0s (0x000000) indicates that the local code is an Ethernet Type II for backward compatibility. This new format moves the Ethernet Type field 8 bytes from its original location in Ethernet II.

Note how the size range for the Data field varies between the different encapsulations (see Table 4-1).

Table 4-1. Size of the Data Field for Different Ethernet Encapsulations

Encapsulation	Minimum Size	Maximum Size
Ethernet II	46	1500
802.3 LLC	43	1497
802.3 SNAP	38	1492

Note that you can send IP datagrams smaller than 46 bytes over Ethernet II because IP contains a Total Length field. When you are sending, for example, 36-byte IP datagrams using Ethernet II encapsulation, a 10-byte trailer with all zeroes is appended to the IP datagram. When you are sending the same IP datagram over 802.3 SNAP, the trailer is only 2 bytes, because 8 bytes are used for the LLC + SNAP headers (1 DSAP + 1 SSAP + 1 Control + 3 OUI + 2 Ethertype).