Ethernet History

Ethernet is the dominant LAN technology at present. It was originally developed at the XEROX PARC (Palo Alto Research Centre) in the 1970s. It was originally designed to work at 3Mbps on a coaxial cable network using CSMA/CD Media Access Control.

It now operates at speeds of up to 10Gbps (10¹⁰ bps) using on a variety of media (co-axial cable, twisted-pair and optical fibre).

1980

XEROX Ethernet (V.1) specification published - based on the famous work of Metcalfe and Boggs. See this paper.

1982

DEC, Intel and XEROX combine to release Ethernet V.2 (DIX Ethernet) specification.

1985

IEEE 802 Technical Committee releases its standard for "Ethernet-like" LANs, 802.3 (pronounced 8 oh 2 dot 3). Virtually all "Ethernet" hardware nowadays complies with the 802.3 standard. Later 802.3 became International Standard ISO 88023

Ethernet over Coaxial Cable - 10Mbps (Obsolete)

The basic idea in early Ethernets was to use a single coaxial cable as the shared communication medium, the Aether or Ether (as "free space" used to be called - when humans thought electromagnetic radiation required something to carry it). Access to the Ether was controlled by the CSMA/CD algorithm (refer to earlier lecture on Media Access Control).

Computers were simply connected to a common cable by "Tee" connectors ("thin coax system").

At the ends of the cable, resistive "terminators" absorbed the signal to prevent reflections.

This scheme used "Thick" (RG8) cable for "backbone" connections. "Thin" (RG58) cable was used to connect the computers together.

Because Ethernet uses CSMA/CD, there were a variety of physical limitations imposed on its operation:

• The "end to end" length of an Ethernet network was limited to about 2.5km by the propagation delay time of the medium and the minimum frame length (64 bytes) adopted by the designers.
• The attenuation of the medium was rigidly controlled so that collisions were unambiguous. The Ethernet specification places strict requirements on the quality and length of a single run of cable (a segment in Ethernet terminology), although the use of repeaters allowed several segments to be connected together.

Ethernet Over Twisted Pair Cable

In this network, the computers are connected to a central "hub" or "switch" by individual "twisted pair" cables which may be up to 100 metres long.

Networks using this technology are variously known as:

• 10BASE-T, CAT3 10Mbps
• 100BASE-T, CAT5 100Mbps
• 1000BASE-T, CAT6 1000Mbps
• 10GBASE-T, CAT6a 10000Mbps

Note: 10BASE-T means:

• 10 million bits per second
• Baseband signalling (no modulation)
• Twisted pair cable

For 10Mbps and 100Mbps Ethernet, two pairs of wires are used (one pair for transmit, one for receive). The connections are made using "RJ45" (8 pin) "phone style" connectors.

Computers and routers use the following pins: 1 - Transmit+ 2 - Transmit- 3 - Receive+ 6 - Receive- Since the wiring is "straight through" (1-1, 2-2, etc) then hubs obviously receive on pins 1 and 2 and transmit on 3 and 6. This means that to connect two similar devices together, a "cross-over cable" is required.

Ethernet Over Fibre

A similar approach to 10BASE-T is taken in the 10BASE-F standards for Ethernet over optical fibre, with two fibres (one for transmit and one for receive) connecting a special transceiver to a 10BASE-F repeater hub.

One excellent advantage of "Ethernet over Fibre" is that a segment can be up to 2000 metres in length.

"Fast" Ethernet

More recently, major new trends have emerged in the Ethernet market:

Switched Ethernet

In ordinary 10BASE-T, the hub is a simple repeater. A switching hub can deliver Ethernet frames from one port to another without other ports being involved. Thus, multiple transmissions through the switch can take place concurrently. Whilst each DTE still operates at 10 Mbps, system wide throughput can be much higher.

100 Mbps Ethernet

In this system, the basic 802.3 technology is simply "scaled up" to 100 Mbps, giving 100BASE-T. Whilst the electronics required to achieve this are rather complicated, the basic idea is still good old CSMA/CD.

Gigabit Ethernet

Scaling the CSMA/CD protocols to 1Gbps has been somewhat more difficult than the increase from 10Mbps to 100Mbps. See later.

CSMA/CD with Hubs and Switches

Behaviour of the newer Ethernet networks differ from the traditional CSMA/CD scheme, depending on the degree of sophistication of the central hub/switch.

Dumb Hub, Half Duplex

• Behaves the same as a coaxial cable network.
• All incoming packets are broadcast on all outgoing circuits.
• Any packet can collide with any other.
• The entire network is said to form a single "collision domain".

Unbuffered Switch

• Packets are forwarded to just one destination (if known to the switch). If the destination is unknown, the packets are broadcast.
• Can handle multiple simultaneous data streams if the sources and destinations are all different.
• The number of collisions are greatly reduced.
• Can still get collisions between packets addressed to a common destination.

Buffered Switch, Full Duplex

• All possible data paths have their own cable/fibre (for both send and receive).
• All data packets are stored before forwarding.
• The switch can delay forwarding packets in order to avoid collisions.
• Collisions are completely eliminated.
• Packets may be lost under heavy load due to the switch running out of storage space.

Ethernet and 802.3 Frame Format

Ethernet is, in general, connectionless, so frames are relatively simple:

Preamble

7 bytes of 10101010

Start Of Frame

1 byte, thus: 10101011

Source and Destination Address

These consists of 3 bytes of "Manufacturer ID", allocated by IEEE, and 3 bytes of "serial number". Thus, every Ethernet interface ever manufactured has a unique "MAC-layer" address. (Thats about 46912 addresses for every person on earth ;-). It's usually written as colon-separated hexadecimal bytes, eg: 08:00:20:0c:a3:55

Type/Length field

If less than or equal to 1500 decimal (0x5DC), then this is the length of an 802.3 packet. If more than 1500, this is the type of an "DIX (DEC/Intel/Xerox) Ethernet V2" packet. For example, 0x0800 indicates that the Ethernet frame contains an IP (Internet Protocol) packet.

Data field

From 46 (preserves CSMA/CD minimum frame size requirement) to 1500 bytes.

Manchester Encoding

The data bits (including preamble and header) are encoded so the resulting electrical signal contains timing information.

Manchester encoding requires a transition in the middle of each bit cell - the direction of the transition (lo -> hi, hi -> lo) indicates logic 1 or logic 0, thus:

Another feature of Manchester encoding is that the average value of the signal can be zero volts, as shown - this allows "AC coupling" of circuits.

Ethernet Protocol Hierarchy

Intro to 802.2 LLC

The IEEE 802 standards committee wanted a protocol-independant method of "de-multiplexing" higher layer protocols encapsulated in various 802-compliant MAC technologies. To allow this, a new Logical Link Control layer was created between the MAC layer and higher layers, thus:

802.2 performs a similar function (in practice) to the "Type" field in a DIX Ethernet frame, except in a technology-independent way.

802.2 LLC Frame Format

The basic 802.2 frame looks like:

DSAP, SSAP are Service Access Point addresses which specify the network layer protocol which generated this frame (SSAP) and the protocol for which it is destined (DSAP). IEEE specifies that 0xFE is used for the OSI network layer.

802.2 defines a variety of services:

Type-1 Connectionless

Operates exactly like Ethernet V.2

Type-2 Acknowledged Connectionless

Received frames are ACK'd, but no initial connection protocol is used.

Type-3 Connection Oriented

Just like High-Level Data Link Control (HDLC)

802.2 SNAP

It became obvious after a short time that the 8-bit DSAP/SSAP field was clearly inadequate for all the possible "higher layer" protocols which might wish to use 802.2 encapsulation.

IEEE solved this problem by defining the Sub Network Access Protocol (SNAP) subset of 802.2 type-1 service.

An 802.2 SNAP frame looks like:

Note that for IP encapsulation in 802.2/SNAP, the "Protocol/Org ID" field is all zeros, and the "Type" field is the standard Ethertype value, 0x0800.

For Appletalk, which uses 802.2 SNAP encapsulation exclusively, the SNAP field is 0x080007809B.

Example Ethernet/802.3 Frame

The following is part of an Ethernet packet captured on the IT student LAN segment at Bendigo, using a packet sniffer running on a Linux based system.

The first 14 bytes of this frame form the Ethernet header comprising:

      6 bytes of destination "MAC" address
      6 bytes of source "MAC" address
      2 bytes of Type/Length

The Type/Length field of 0x00cd (hexadecimal) = 205 decimal (less than 1500 decimal), tells us this is an "802.3" packet.

An "802.3" packet always contains a Link Level Control (LLC) header.

The LLC header indicates (by the bytes aa aa) that the 802.3 packet contains an SNAP header.

The SNAP header, in turn, indicates that the 802.3 packet contains an Appletalk packet - (probably having its own header).

Note that each header tells us what follows in the packet.

References

Stallings: Chapter 16 (pp. 442-481); Chapter 15 (pp. 482-521)