Robert Metcalfe deserves credit for conceiving and developing Ethernet. It was his insight and persistence that created a truly universal protocol and hardware substrate. Token Ring, its rock solid theoretical basis not withstanding, could never aspire to Ethernet's universality so that it never really had a chance.

Ethernet was not developed in a vacuum. Other network connectivity systems were already in use. One of these, Aloha, was fully functional before Ethernet got off the ground. It is worth taking a look at this wireless setup in order to fully appreciate the birth and early development of Ethernet. Also called "Alohanet," this network was invented by Norman Abramson and his associates at the university of Hawaii in 1970. Like Ethernet, Aloha was developed in response to a specific need.

The University of Hawaii consisted of multiple, widely separated campuses. It was a natural solution to use amateur radio to link their computers. Thus wireless came first. A star topology with a central hub was created. Two frequencies were employed, one to transmit and one to receive. A machine that received data immediately retransmitted it so that comparison would reveal any corruption, in which case after a short wait the packet would be retransmitted. It was possible to compensate for data collision when more than one machine tried to send packets simultaneously.

The distinguishing characteristic of this network was its use of shared media. Since the Data Terminal Equipment (DTE) was 80 words per minute Teletype, speed was not of the essence. The whole issue of data collision detection was not too important, so Aloha essentially finessed it. If a collision occurred, both transmissions failed and the data would be resent manually. At its characteristic low speed, the network functioned satisfactorily without an automatic collision solution. Robert Metcalfe had his opening.

Ethernet, as we shall see, came into existence in 1973, more or less contemporaneous with Token Ring. Both systems dealt effectively with data collision albeit in totally different ways. The demise of Token Ring was not at all due to inability to deal with collisions, but rather for other reasons. Chief among these was the sheer elegance and universality of Ethernet, which is why today it is the dominant networking technology comprising over 85% of an enormous worldwide market.

Token Ring was developed by IBM and came on very strong in the 1970's, even as Ethernet was getting off the ground. How does Token Ring solve the data packet collision problem? The idea is simple. Each LAN (Local Area Network) Token Ring DTE element occupies a position within ring topology. Data is transmitted from station to station accompanied by a control "token" permitting access. Whoever possesses the token can talk. Everyone else has to wait. Needless to say, this token is not a physical object made of plastic or metal. It has a virtual existence in cyberspace. Specifically, it is a three-byte frame. You cannot transmit your data frame until you have acquired the token frame.

Other proprietary token schemes were in existence. Apollo Computer came out with a 12 Mb/s methodology in 1981 and Proteon offered its 10 Mb/s ProNet 10 Token Ring network in 1984. The pre-eminent example of this technology was IBM's 16 Mb/s unit introduced in 1989. This network was made into the IEEE standard.

Later, a speed of 100 Mb/s was achieved, but by this time Ethernet eclipsed Token Ring, because of the latter's unwieldy physical substrate. The cabling was IBM Type 1, shielded twisted pair with bulky fragile plastic connectors. Even though data collision had been dealt with, these rival systems never came close to Ethernet's speed, universality and simplicity.

Prior to the early 1970's, it was presumed that the future of computers would be larger mainframes, which would occupy large buildings and require enormous water-cooled physical plants.

An important innovation at this time was the personal computer. Rather than gigantic sentient structures, the idea of small discrete workstations with central processing units attached to monitors and keyboards (the mouse was as yet undreamed) gained ascendancy.

Palo Alto Research Center (PARC) owned and operated by Xerox, was one site of this remarkable development. Robert Metcalfe, a member of the research staff, was charged with the task of networking PARC's computers. Xerox had just invented the laser printer and wanted all of the computers (there were hundreds in one building) to access the printer at very high speed.

May 22, 1973, can be taken as the birth date of Ethernet. On that day Robert Metcalfe outlined to company management the new protocol. Several year's work was necessary to make it a reality. A milestone came in 1976 when Metcalfe and his associate, David Boggs, published and article laying it all out, "Ethernet: Distributed Packet-Switching for Local Computer Networks." What a wonderful name! It is reminiscent of the theory in classical physics that an intangible but nevertheless real substance, Ether, permeated the entire universe and constituted the medium for the propagation of light.

The original Ethernet networks employed a serial bus topology. Coaxial cable and other off-the-shelf elements comprised the physical layer. At the heart of the Ethernet protocol was CSMA/CD, which is an acronym for Carrier Sense Multiple Access Collision Detection. This means each station continuously listens for traffic on the medium to determine if the current time slot is vacant. Any time that conditions is met, any or all stations may transmit. If two or more stations begin transmitting at the same time, data packet collision will occur, all such transmission will be unreadable, and they must be aborted. Then, in response to a particular algory thm, each transmitting station waits a quasi-random period of time and then retransmits.

This scheme worked well at slower speeds but the need for faster speeds arose and new developments surfaced. This is why Ethernet predominated -- it possessed a remarkable flexibility that allowed it to reconstitute itself at ever-higher levels. Some of the developments included:

• Serial bus topology was replaced by star topology. Individual nodes are connected to a central hub so that in any given path there are fewer terminations that might compromise network speed and accuracy. More cable was needed, but that was not too big an issue.
• Full duplex transmission allows simultaneous two-way communication. There is no media contention or collisions to worry about so that greater efficiency is attained.

  VLAN tagging was developed to implement network prioritization. A VLAN tagged frame is a MAC data frame that has a four-byte header, which consists of two fields, a two-byte indicator that the frame is a VLAN frame and another two-byte tag-control field that contains the transmission priority, 0 to 7.

  The medium changed from coaxial cable (first Thicknet and then the easier to work with Thinnet) to UTP (unshielded twisted pair) also known as category cable.

  In 1979 Robert Metcalfe, having left Xerox, set out to further develop the idea of a LAN composed of many small personal computers. It was clear that the success of Ethernet depended upon putting an end to the proliferation or proprietary protocols with incompatible hardware. He presented the idea of Ethernet as an industry standard to three major players in the computer arena: Digital Equipment corporation, Intel and Xerox. The first Ethernet specification was issued in September, 1980. It was titled "DIX" from Digital, Intel and Xerox, and it was also called the Ethernet Blue Book. In 1982 IEEE formed Project 802 to codify the emerging standard. A subcommittee, 802.3, created the international document, which was subsequently, in 1985, adopted by ISO (International Standards Organization). The great achievement was the development of an open system with cross-platform and multi-vendor standards. Because of these moves, Ethernet became very user friendly, inexpensive and reliable.

  At this point in time, Ethernet ran at 10 Mb/s. It employed a large yellow coaxial backbone cable with smaller coaxial cables tapped off to connect individual workstations. This so-called "thick Ethernet" was denominated as 10Base-5 (the 10 indicating the system speed and the 5 referring to the maximum cable length of 500 meters).

  In 1985, IEEE released Version Two, which became known as thin Ethernet of 10Base-2. The 2 is shorthand for its maximum length, actually 185 meters.

  Subsequent Ethernet incarnations employed a vastly improved medium, or cable. It is UTP and recent Ethernet history is made up of a series of improved versions of the Category cable:

  • Category 1 is suitable for only telephone and ISDN work. It is not twisted and capable of a maximum data rate of only 1 Mb/s.
  • Category 2 has a maximum data rate of 4 Mb/s and found use in Token Ring.
  • Category 3 has a maximum data rate of 10 Mb/s and was also used for Token Ring.
  • Category 4 has a maximum data rate of 16 MB/s and was also used in Token Ring.
  • With the introduction of Category 5, the data rate jumped up to 100 Mb/s, achieving the higher speed because it has more twists per inch.
  • Category 5e replaced Category 5 as a standard and achieved a data rate of 1000 Mb/s and is of great use today since it can be the medium for Gigabit Ethernet.
  • Category 6 and 6a (augmented) are even better because they can run 10 Gigabit Ethernet, but due to price and limited distance they are inviting a takeover by the increasingly attractive optical fiber option.

  UTP cable is terminated using an RJ-45 connector. Preterminated patchcords using stranded cable are readily available and longer runs of solid cable can be terminated in the field using a cable stripper and RJ-45 crimper tool. Stranded and solid cable each has its own type of connector and it is important to use the correct one.

  Undoubtedly optical fiber or wireless (often in conjunction with optical fiber) will eventually predominate, but for now Category 5e and 6 are the big players. Category 5e is easy to install and terminate and appropriate to commonly encountered network speeds.

  A relatively recent addition to the Ethernet family has been Power of Ethernet (PoE). This innovation allows for a DC voltage to be applied to the UTP cable for the purpose of powering a device where regular AC power is not available at the location. PoE is frequently used for powering wireless LAN access points, IP telephones, security cameras, embedded computers and remote network switches. Because this technology may be applied to a network at some time in the future, it is crucial that unused twisted pairs not be miswired or damaged in a way that might apply voltage where it is not wanted. To power up one of these systems, a Category 5e injector inserts the DC voltage. This injector is often located at the source adjacent to the Ethernet hub or switch. PoE compatible devices pick up the power directly through the RJ-45 jack.

  The current IEEE PoE standard, 802.3af, calls for 48 volts DC, but it is important to realize that various vendors utilize different PoE standards and only compatible equipment may be used.

  The success of Ethernet has been due to its ability to combine with new technologies such as optical fiber as they have become available. Industrial Ethernet has made its way into PLC (Programmable Logic Controllers in the manufacturing facility and other applications are on the horizon.
  --END--