Bitzenbytes.com

Ethernet, Round 1

Most home networks employ some form of Ethernet, or descendant network technology, as a basis for data transmission between computers.   The data is, not uncommonly, formed according to TCP/IP protocols under some kind of network architecture, typically one of the Windows network architectures.  TCP/IP is the default network communications protocol for Windows, and er, Ethernet is the most practical media specification that will support TCP/IP.  Some home networks use AppleTalk Network Architecture, still on Ethernet.  Others may use something else.  In fact, these statements are true of many business networks as well, albeit there are very large chunks of business that employ other major network architectures.

It is becoming increasingly common to see wireless extensions added to a basic home network, so some wireless networking terminology, though not Ethernet, will be included in this document.  Ease of deployment may help wireless home networks become the majority of home network installations at some not-so-distant future time.

Wireless technology for home networks is still in its infancy and has a long way to go.   While easy to deploy, it offers less bandwidth, less security, and less dependability when compared to cable installations.  As bandwidth needs increase, wireless technology may not be able to keep up.  The convenience of wireless technology, however, ensures that it will have a place among networks now and in the future.

Not that this matters too much to the home networking enthusiast.  Fortunately, the same network technology, used in the single family home, was initially scaled to fit the demands of office buildings.  It will be the unusual house that can't centrally locate the network "backbone" device, and the unusual home network that can't be contained in a 100 meter (328 feet--the football field with end zones) radius of that device.  Even if the "backbone" device is not centrally located between the two football fields in your house, and is hung on one wall, your computers will probably be located inside a 100 meter swing from that location.  Even if central location wasn't enough to cover the field, the ambitious owner of such a sprawling mansion could employ a relatively cheap router (and some amount of additional complication) and push the distance out another 200 meters on copper.  [Eah, I will be happy to consult with anybody who has a house with wire runs longer than 640 feet (we'll be using fiber in your house anyway.)]

Even for wireless, distance should not be a problem in most houses (provided the construction of the house does not interfere with transmission.)  While Ethernet, business networking technology and home networking technology have seen huge increases in bandwidth, large increases in distance, and gigundus decreases in cost, almost any of the Ethernet technologies of the past 25 years would still support today's modest home network requirements.

Modest Home Network Configurations

To successfully build a local area network design, it is necessary to understand, at a basic level, what affects network function and how the network performs at the first three Open Systems Interconnection (OSI) layers.  The current market standard for these layers is:

• Internet Protocol (IP) for the networking layer
• Ethernet for the data link layer
• Unshielded Twisted Pair (UTP) copper for the physical layer

To understand network addressing and communications at OSI layer 3, you will need a basic knowledge of:

• binary and hexadecimal arithmetic
• the basis of IP addressing
• how to subnet a local network segment
• how the TCP/IP protocol suite functions
• how the Domain Name System (DNS) works
• how networks employ routing to deliver messages
• Network Address Translation (NAT), Ports, Firewalls

Ethernet technologies are thought to be the realm of the IEEE 802.3 specification.   Some of the LAN technologies covered by 802.3, however, bear little resemblance to the original Ethernet specifications.  New signaling methods are employed, CSMA/CD is made unnecessary, the media is changed, new methods of isolating traffic are introduced, and little remains of the original specification.  Packet format is consistent along with some restrictions that still allows earlier Ethernet equipment to be used with the new.  Little seems to change at the Data Link layer while newer equipment at the physical layer is accommodated.

• format or structure of data
• method of error checking
• method of data compression
• required control information
• method of indicating that a message is finished
• method of indicating that a message is received
• other required information

One mnemonic phrase used to remember the names of the layers is   "All People Seem To Need Data Processing."  That's seven words with seven leading letters to recall Application, Presentation, Session, Transmission, Network, Data Link, and Physical... ...the seven OSI layers in correct descending order.  While this document will discuss the first two layers, Physical and Data Link layers, it will help to be familiar with the others.

The OSI model is not a specification but is a generally descriptive model for a Network Architecture.  By comparing a given network  to the model, we are able to describe and compare diverse Network Architectures that are fleshed out with real protocols and real implementations.  The model is a framework for discussion.  Without the model, chaos tends to reign.

The OSI model was created to meet some basic organizational and design requirements.  Each layer of the OSI model is designed so that it performs a well defined function where that function can be clearly exposed by a set of standardized protocols.   Layers are established to accommodate different levels of data abstraction.   Layers should be as small as possible for the elegance and ease-of-use delivered by simplicity, and should be large enough to accommodate the functional requirements of the layer.  Layer boundaries are established at points where a minimum flow of data must pass the boundary.   Boundaries are sometimes established by programming interfaces.

Comparison of IEEE Project 802 Model Layers with OSI Model Layers

Simply put, however, the layers of the OSI  model are like links in a chain.  Break any of the links and network communications at some level will fail.  Each layer communicates through Protocol Data Units (PDU).

In general, when you see references to layer 1, layer 2, and layer 3 technology, these are, respectively, references to Physical, Data Link, and Network layers of the OSI model.  This document is concerned primarily with the Physical layer and the Datalink layer.

Bob Metcalfe left Xerox in 1979 to found 3COM Corporation and 3COM shipped its first Ethernet product in 1981.

The original Ethernet specification, published by Digital, Intel, and Xerox (DIX) in 1980 was eventually superceded by IEEE 802.3 in 1985 when it became an open industry standard and was described as Carrier Sense Multiple Access with Collision Detection (CSMA/CD).  The original specification was called Ethernet and this was followed by Ethernet II.  Then followed the IEEE specification.  However, the IEEE essentially adopted most of the Ethernet specification for 802.3.  Everybody still calls it Ethernet, but the spec is 802.3 and others depending on implementation.    IEEE 802.3 is periodically updated to include new network technology.    IEEE P802 is the LAN/MAN Information Technology working group or Standards Committee.   Look for the letters that follow 802.3_

In general, Ethernet devices attach to a common transmission medium.  The medium provides a transmission path where electronic signals can travel.  Initially, for "classic" Ethernet, this medium has been coaxial copper cable. Coax is gradually declining in use, and today twisted pair copper or fiber optic cabling is more commonly used. Primarily, this is due to the cost of coaxial cable, and its limited potential when compared to fiber.  Regardless, any single shared medium is called an Ethernet segment, and computer devices that attach to that segment are stations or nodes. In the case of wireless transmission, the segment is defined as the maximum distance at which successful propagation between any station and an access point can occur.

In 1983, 3COM shipped its first NIC (Network Interface Card) for the IBM-PC (and IBM-AT) using RG-58 thinwire coax cable and BNC connectors (10BASE-2 -- 180 meters.)  DEC was delivering devices that used AUI cable between a NIC and a spinal tap transceiver on RG-8 thick wire (10BASE-5 -- 500 meters) using 50 ohm terminators and N connectors.    DEC soon entered the "light weight"  market with RG-58 specifications and a number of devices that helped with deployment on existing networks including a multi-port repeater (DEMPR) that connected 8 coax segments.   ...and that's approximately where I entered the game as a small business systems administrator who pulled his own thickwire coax and made his own cable taps.  This was the same year that Xerox gave away the Ethernet patents to the IEEE which licenses any company to build Ethernet hardware for a small fee (you could afford the fee.)

In 1985, the IEEE published the basic thick (RG-8) coaxial Ethernet specifications.   The open 10 Mbps Ethernet standard provided a communications path for heterogeneous networks, and for its time provided a flexible and inexpensive method of network implementation.

The ISO accepted Ethernet as a standard in 1989 (Standard # 88023) -- strange similarity in the numbers?

In 1990, 10BASE-T (IEEE 802.3i) worked. ...and now anybody can do 100BASE-TX (IEEE 802.3u) at home for cheap. In 2002, we're all waiting for gigabit Ethernet (1000BASE-T) to come down in price. ...and others are waiting for the wireless arm of the industry to standardize on a bug-free, working security standard and a decent rate of transmission.

Because Metcalfe, Boggs, Xerox, and the IEEE essentially gave the knowledge and specifications of the Ethernet invention away for free, the standard is now ubiquitous all over the world.

At one time, installing a NIC for 10BASE-T required some significant amount of PC knowledge including how and where to reserve upper memory, what IRQ to allocate and what ports to assign.  The mantra for people installing NIC's was "NE2000", and the nemesis was "RAM-cram."  NE2000 was a Novell NetWare specification for an Ethernet NIC that became so much used that just about anything having to do with PC LAN's had to be compatible with the operation of a NE2000 NIC.  NE2000 compliant clones were common.   NIC's and OS's and installation software have gotten smarter.    Now, it's plug it in, let the OS recognize it and feed it a device driver.

TCP/IP is not dependent on Ethernet for a physical layer, but it is the most commonly implemented physical method used for TCP/IP transmission on a LAN.  It is almost the only method used for home networking with most other methods being more costly and subject to the acquisition of equipment that is less than available to John Q. Netizen.  TCP/IP works with Ethernet to provide network services.  IP works at the OSI model Network layer which is the layer just above the Data Link Layer.

Ethernet is both an OSI Model Physical layer and Data Link layer specification.    Just note that the physical layer of the OSI model is not restricted to Ethernet.  TCP/IP can be transmitted over networks with other physical specifications such as Token Ring, FDDI, various wireless technologies (including 802.11) and others.

The physical address is the MAC address.  The MAC address is generally burned into a chip on the network interface card (NIC) or other network adapter device.  The MAC address is a 48 bit binary address that is notated as a 6 digit hexadecimal address--where the digits are separated by dashes.  The first three digits designate a manufacturer code for the device and the second three digits are a unique address for the device on a given network segment.

Fortunately, MAC addresses need only be unique on a local network segment (subnet.)    Any limitations, that might be imposed by the fact that there are a finite number of MAC addresses, should be somewhat easy to escape.

The data packet that Ethernet receives from IP is encapsulated with a header and a trailer at Data Link layer.  The packet is then transferred as a signal on the Ethernet medium, typically copper or fiber.

The physical components of most networks are specified as being part of the OSI model physical and data link layers.  Such specifications include specifications for the encoding of data, the attributes of the signaling used, the electrical properties of the equipment, cabling, the methods used to connect cabling, and devices that operate at physical layer such as repeaters, hubs, and switches. 

In general, all of these specifications for equipment, transmission type, and protocols contribute to a construct that is referred to as Network Architecture.  Network Architecture is the totality of the implementation of these specifications that permits a network to operate.  Some examples of network architectures are SNA, DECNet, XNS, Banyan Vines, Apple Talk,  and TOPS.  These network architectures incorporate network protocols (such as NetBEUI, IPX/SPX, and TCP/IP).   Modern network architectures can often be explained in terms of the OSI reference model.    For each one of these network architectures, you can find a full set of specifications that will enable communications across a network to be accomplished.    Each one of these architectures implements a protocol stack ( a series or set of rules that work together) that allows programmers to create programs that use and alter data housed and manipulated on other computers.

It is at data link layer that data is prepared to be transmitted and received on the physical layer.  Data link layer also provides for some error detection.  Though Ethernet encompasses both physical and data link layers, the tasks performed can be organized according to a comparison to the OSI model.

For Ethernet, at OSI model physical layer, there are specifications for

• cable
• connectors
• topology (methods for interconnection)
• encoding of data
• signaling
• electrical attributes of equipment
• physical layer equipment

for Ethernet, at OSI model Data Link Layer, there are specifications for

• Media Access Control (physical) Addresses for network devices
• method for framing an IP packets in datagrams (data frames)
• transmission of data bits from the computer system to the NIC
• Media Access Control (NIC to medium)
• Logical Link Control (LLC) error detection

The Ethernet Physical layer of any given network could have several different specifications depending on media type.  Essentially, the physical layer will depend on three parts of the Ethernet specification:

• The shared media
• A datagram or Ethernet frame
• A set of media access control rules embedded in the NIC

Any computer, or other intelligent device, with a network address, and located on a network is called a host.  A host with a network interface that includes a physical network address can be called a station.  Every station on an Ethernet network segment shares the bandwidth available on the media.  This shared signaling system is called the medium.  Ethernet signals, consisting of a datagram or packet, are transmitted serially, one bit at a time.  When one station is talking, all of the other stations must listen.  To send data, an Ethernet station must listen to the "wire" segment and if the "wire" segment is quiet then the station may start its send sequence.

Please note that the links in the article above were checked within the last two weeks. I note that several may have decayed already. Web sites frequently change the location of their documents or remove documents that may not be accessed often. If a link above takes you to a site, but the expected document is not displayed, try searching the site for the document--they may just have moved it.

Alternately, if you take the term that was linked and submit that string to a search engine, it is likely that you will find numerous references to any topic in the post above.

More about Ethernet in Round 2.