Overview of the Ethernet Protocol and its types

Ethernet is one of the oldest LAN technology but still the most popular and the most widely used one in LAN environments. It is used in almost all networks including home, office, universities and enterprise networks. Of late, due to its higher rates over longer distances through optical media, it has gained popularity also as a WAN technology too.

Ethernet is basically a

• next hop carrier data link layer protocol for carrying network layer packets between adjacent hops
• a connectionless best effort data link layer protocol. That means it does not have the LLC layer functionality of logical links and applications should rely on the higher layers to provide reliability functionality.
• For shared lines (e.g. bus), Ethernet supports the media access control (MAC) functionality through the use of the standard CSMA/CD protocol.
• It supports error detection through CRC and flow control through PAUSE frames.
• It does not support reliability and error correction
• Its powerfulness comes due to its support for a wide variety of physical layer protocols, physical media, line rates and topologies (bus, star, point to point).
• Its wide adaptability comes from its simple nature (connectionless, best-effort delivery approach) , backward compatibility between different variants (like FE, GigE etc.) and cheap cost of hardware implementation.
• It supports half and full duplex mode of communication, though half duplex mode is outdated now.

Basic Frame Structure of Ethernet

Ethernets framing structure follows the general data link layer frame structure, with fields for frame boundaries, addressing, protocol multiplexing, data and error detection (CRC). A typical Ethernet frame format is given in the diagram below:

A typical Ethernet Frame

As shown in the diagram, an Ethernet frame consists of the following fields:

• A 7 byte repeated pattern of the value 10101010 named as Preamble, used to synchronize the receiver’s clock with that of the sender’s clock
• A 1 byte Start-Of-Frame with value 10101011, to indicate to the receiver that the actual frame fields starts from the next byte after this special byte.
• A 6 byte destination MAC address identifying the destination node of the frame at the data link layer
• A 6 byte source MAC address identifying the sending node of the frame at the data link layer
• An optional 4 byte VLAN header tag identifying the VLAN to which the frame belongs. This field is optional and is present only if VLANs are used in the network.
• A 2 byte protocol type field containing the type of the layer 3 packet contained in the frame
• A variable length DATA field that contains the actual L3 payload. The maximum length of this field is limited to 1500 bytes for normal ethernet frames, but can go upto higher values (around 9000 bytes) for special frames known as jumbo frames.
• An optional padding field filled with value all zeros, used mainly to maintain a minimum Ethernet frame length. This is used only in cases where the DATA field is lesser than 46 bytes. An Ethernet frame has to be of minimum length equal to 64 bytes, including the 18 byte Ethernet header.
• A 4 byte CRC field used for error detection purposes.
• A normal Ethernet frame header is 18 bytes long, including the 4 byte CRC. 

Variants of Ethernet

There are different variants of Ethernet like 10Mbps Ethernet,100Mbps ( Fast Ethernet -FE),1000Mbps ( Gigabit Ethernet -GigE), 10 GigE, 100 GigE etc., mainly based on the data rate. WIthin each variant, there are multiple sub-types like 10Base2, 10BaseT, 1000BaseT etc., based on the type of the underlying physical media, the cable type, the maximum distance supported etc.

The following hold true for the different variants of Ethernet:

• Framing structure remains the same across all variants, thereby enabling backward compatibility and allowing connectivity between Ethernet interfaces with different maximum speeds.
• Main difference comes at the physical layer based on the following attributes
  • Maximum interface speed (10Mbps, 100 Mbps, 1000 Mbps, 1000Mbps, 10000Mbps etc.)
  • Line Encoding Techniques (e.g. Manchester encoding, 4B/5B encoding, 8B/10B encoding etc.)
  • Media types used (it can either be copper or optical)
  • Cable Types & Connectors (E.g. Cat2, Cat5, Cat 6 etc.)
  • Max Distance Separating nodes in the network (e.g. 10 metres, 20 meters, 100 meters etc.)
• MAC is based on the CSMA/CD protocol

CSMA/CD

In all the variants of the Ethernet, CSMA/CD is the standard media access control collision resolution protocol used. CSMA/CD  is used mainly in bus topologies and in half duplex modes, where multiple nodes share the same bus and thereby contend for a channel. Basically CSMA/CD is based on:

 CSMA (Carrier Sense Multiple Access) – Each end node starts transmitting only if the line is not busy already, by looking for the presence of a carrier signal

 CD ( Collision Detection) – If the value of the signal voltage  transmitted on the line by a node is different from the line voltage appearing on the line, then the end node detects this as a collision. It immediately sends a special signal called as a JAM signal and stops further transmission of the current frame. Once a collision is detected, an end node uses an exponential backoff algorithm before it attempts retransmission.