Ethernet

format_list_bulleted Contenido keyboard_arrow_down
ImprimirCitar
10 Mbit/s ISA Network Card.
BNC (Coaxial) and RJ45 connectors from a network card.
UTP cable for Ethernet connections.

Ethernet (English pronunciation: /ˈiːθə˞nɛt/) is a local area network standard for computers, for its acronym in Spanish Multiple Access with Carrier Listening and Collision Detection (CSMA/CD). Its name comes from the physical concept of ether (ether, in English). Ethernet defines the characteristics of cabling and signaling; physical layer and the data frame formats of the data link layer of the OSI model. Ethernet was taken as the basis for the drafting of the international standard IEEE 802.3, being usually taken as synonyms. They differ in one of the fields of the data frame. However, Ethernet and IEEE 802.3 frames can coexist in the same zone.

History

In 1970, while Norman Abramson was setting up the great ALOHA network in Hawaii, a recent MIT graduate student named Robert Metcalfe was doing his doctoral studies at Harvard University working for the ARPANET, which was the hot topic of research in those days. On a trip to Washington, Metcalfe was at the home of Steve Crocker (the inventor of Internet RFCs) where he let him sleep on the couch. In order to fall asleep, Metcalfe began to read a scientific magazine where he found an article by Norm Abramson about the Aloha network. Metcalfe thought about how the protocol used by Abramson could be improved, and wrote an article describing a protocol that substantially improved the performance of Aloha. That article would become his doctoral thesis, which he presented in 1973. The basic idea was very simple: stations before transmitting should detect if the channel was already in use (that is, if there was already a 'carrier')., in which case they would wait for the active station to end. In addition, each station while transmitting would be continuously monitoring the physical medium in case of a collision, in which case it would stop and retransmit later. This MAC protocol would later be called Carrier Sense Multiple Access and Collision Detection, or CSMA/CD for short (Carrier Sense Multiple Access / Collision Detection).

In 1972 Metcalfe moved to California to work at the Xerox Research Center in Palo Alto called Xerox PARC (Palo Alto Research Center). There he was designing what was considered the & # 39; office of the future & # 39; and Metcalfe found a perfect environment to develop his concerns. Some computers called Alto were being tested, which already had graphics capabilities and a mouse and were considered the first personal computers. The first laser printers were also being manufactured. They wanted to connect computers together to share files and printers. The communication had to be very high speed, in the order of megabits per second, since the amount of information to be sent to the printers was enormous (they had a resolution and speed comparable to a current laser printer). These ideas that seem obvious today were completely revolutionary in 1973.

Metcalfe, the team's 27-year-old communications specialist, was tasked with designing and building the network that would tie it all together. He had the help of a Stanford doctoral student named David Boggs. The first experiences of the network, which they called the 'Alto Aloha Network', were carried out in 1972. They gradually improved the prototype until on May 22, 1973, when Metcalfe wrote an internal memo in which he informed of the new network. To prevent it from being thought that it was only used to connect computers, Alto changed the name of the network to Ethernet, which made reference to the now-abandoned theory of physics according to which electromagnetic waves traveled by a fluid called ether that was supposed to fill all space (for Metcalfe the 'ether' was the coaxial cable through which the signal went). The two Alto computers used for the first Ethernet tests were renamed Michelson and Morley, after the two physicists who demonstrated in 1887 the non-existence of the ether through the famous experiment of Michelson and Morley.

The 1973 network already had all the essential features of today's Ethernet. It used CSMA/CD to minimize the probability of a collision, and in the event of a collision, a mechanism called binary exponential regression was put into operation to gradually reduce the 'aggressiveness' of the emitter, with which it adapted to situations of very diverse traffic level. It had a bus topology and operated at 2.94 Mb/s over a 1.6 km long segment of coaxial cable. The addresses were 8 bits and the CRC of the frames 16 bits. The protocol used at the network level was the PUP (Parc Universal Packet) which would later evolve into what was later XNS (Xerox Network System), predecessor of IPX (Novell Netware).

Instead of using the 75 ohm coaxial cable of cable television networks, it was decided to use 50 ohm cable that produced less reflections of the signal, to which Ethernet was very sensitive for transmitting the signal in base band (ie without modulation). Every cable splice and every 'spike' installed vampire (transceiver) produced the reflection of a part of the transmitted signal. In practice the maximum number of 'skewers' Vampire, and therefore the maximum number of stations on a coaxial cable segment, was limited by the maximum tolerable reflected signal strength.

In 1975 Metcalfe and Boggs described Ethernet in an article they sent to the Communications of the ACM (Association for Computing Machinery), published in 1976. In it they already described the use of repeaters to increase the range of the network. In 1977 Metcalfe, Boggs, and two other Xerox engineers received a patent for the basic Ethernet technology, and in 1978 Metcalfe and Boggs received another for the repeater. At this time the entire Ethernet system was owned by Xerox.

It should be noted that David Boggs built the first Internet router and name server in 1975 during his stay at Xerox PARC.

Versions of 802.3

Ethernet was taken as the basis for the drafting of the international standard IEEE 802.3, being usually taken as synonyms. They differ in one of the fields of the data frame. However, the original Ethernet and IEEE 802.3 frames can coexist on the same network.

The standards of this group do not necessarily reflect what is used in practice, although unlike other groups this one is usually close to reality.

The first version of IEEE 802.3 was an attempt to standardize ethernet although there was a header field that was defined differently, there have been subsequent successive extensions to the standard that covered speed extensions (Fast Ethernet, Gigabit Ethernet and the 10 Gigabits), virtual networks, hubs, switches and different types of media, both fiber optic and copper cables (both twisted pair and coaxial).

Standard Ethernet Date Description
Experimental Ethernet 1972 (paid in 1978) 2,85 Mbit/s on coaxial cable in bus topology.
Ethernet II (DIX v2.0) 1982 10 Mbit/s on fine coaxial (thinnet) - The plot has a package type field. The IP protocol uses this plot format over any medium.
IEEE 802.3 1983 10BASE5 10 Mbit/s on coaxial thick (thicknet). Maximum length of the segment 500 meters - Just like DIX except that the Type field is replaced by the length.
802.3a 1985 10BASE2 10 Mbit/s on fine coaxial (thinnet or cheapernet). Maximum length of the segment 185 meters
802.3b 1985 10BROAD36
802.3c 1985 Specification of 10 Mbit/s repeaters
802.3d 1987 FOIRL (Fiber-Optic Inter-Repeater Link) fiber optic link between repeaters.
802.3e 1987 1BASE5 or StarLAN
802.3i 1990 10BASE-T 10 Mbit/s on non-armoured braided pair (Unshielded Twisted Pair o UTP). Maximum length of the segment 150 meters.
802.3j 1993 10BASE-F 10 Mbit/s on fiber optics. Maximum length of segment 1000 meters.
802.3u 1995 100BASE-TX, 100BASE-T4, 100BASE-FX Fast Ethernet at 100 Mbit/s with speed self-negotiation.
802.3x 1997 Full Duplex (Simultaneous transmission and reception) and flow control.
802.3y 1998 100BASE-T2 100 Mbit/s on non-armoured braided pair (UTP). Maximum length of segment 100 meters
802.3z 1998 1000BASE-X 1 Gbit/s Ethernet on fiber optics.
802.3ab 1999 1000BASE-T Ethernet 1 Gbit/s over non-armoured braided pair
802.3ac 1999 Maximum plot extension to 1522 bytes (to allow the "Q-tag") Q-tags include information for 802.1Q VLAN and manage priorities according to standard 802.1p.
802.3ad 2000 Adding parallel links.
802.3ae 2003 Ethernet to 10 Gbit/s; 10GBASE-SR, 10GBASE-LR
IEEE 802.3af 2003 Power over Ethernet (PoE).
802.3ah 2004 Ethernet in the last mile.
802.3ak 2004 10GBASE-CX4 Ethernet at 10 Gbit/s on bi-axial cable.
802.3an 2006 10GBASE-T Ethernet at 10 Gbit/s on non-arm braided pair (UTP)
802.3ap in process (draft) 1 and 10 Gbit/s Ethernet on printed circuit.
802.3aq in process (draft) 10GBASE-LRM Ethernet at 10 Gbit/s on multimode optical fiber.
802.3ar in process (draft) Congestion management
802.3as in process (draft) Extension of the plot

Ethernet frame format

The plot is what is also known as the "frame".

Payload structure on Ethernet and IP and TCP protocols.
  • The first field is the preamble that indicates the beginning of the plot and has the object that the device receiving it detects a new plot and syncs.
  • The plot start delimiter indicates that the frame starts from it.
  • The MAC fields of destination and origin indicate the physical addresses of the device to which the data and the source device of the data are directed, respectively.
  • The label is an optional field that indicates membership of a VLAN or priority at IEEE P802.1p
  • Ethernetype indicates with which protocol the data contained in the Payload are encapsulated, in case a higher layer protocol is used.
  • The Payload is where all the data go and, if any, headers of other higher layer protocols (According to OSI Model, see Computer Protocols) that could format the data that are processed (IP, TCP, etc.). It has a minimum of 46 Bytes (or 42 if it is version 802.1Q) up to a maximum of 1500 Bytes. Messages below 64 bytes are called dwarf plots (runt frames) and indicate damaged and partially transmitted messages.
  • The check sequence is a field of 4 bytes that contains a CRC verification value (cicle redundancy control). The emitter calculates the CRC of the entire plot, from the field to the field CRC assuming that it is worth 0. The receiver recalculates it, if the calculated value is 0 the plot is valid.
  • The frame end gap is 12 empty bytes with the goal of spacing between plots.
802.3 Ethernet frame
PreambleDelimiter of start of plotTarget MACMAC of origin802.1Q Tag(optional)Ethertype (Ethernet II) or length (IEEE 802.3)PayloadSequence of evidence (32-bit CRC)Gap between frames
7 Bytes1 Byte6 Byte6 Bytes(4 Bytes)2 BytesFrom 46 (or 42) to 1500 Bytes4 Bytes12 Bytes
64–1522 Bytes
72-1530 Bytes
84–1542 Bytes

Ethernet Technology and Speed

Ethernet has long since established itself as the primary link layer protocol. Ethernet 10Base2 already achieved wide acceptance in the sector in the 1990s. Today, 10Base2 is considered a "legacy technology" compared to 100BaseT. Today manufacturers have already developed adapters capable of working with both 10baseT and 100BaseT technology and this helps to better adaptation and transition.

The Ethernet technologies that exist differ in these concepts between them:

Transmission speed
Speed to which technology transmits.
Cable type
Physical level technology used by technology.
Maximum length
Maximum distance between two adjacent nodes (without repeating stations).
Topology
Determine the physical form of the Bus network if T connectors are used (today only used with older technologies) and star if used hubs (dissemination star) or switches (commuted star).

The above concepts are specified below in the most important technologies:

Ethernet technologies
TechnologyTransmission speedCable typeMaximum distanceTopology
10Base5 10 Mbit/s Coaxial thick 500 m Bus (AUI Connector)
10Base210 Mbit/sCoaxial thin185 mBus (Conector T)
10BaseT10 Mbit/sPaired100 mStar (Hub or Switch)
10BaseF10 Mbit/sOptical fibre2000 mStar (Hub or Switch)
100BaseT4100 Mbit/sBraided Par (category 3UTP)100 mStar. Half Duplex (hub) and Full Duplex (switch)
100BaseTX100 Mbit/sBraided (category 5UTP)100 mStar. Half Duplex (hub) and Full Duplex (switch)
100BaseFX100 Mbit/sOptical fibre2000 mIt does not allow the use of hubs
1000BaseT 1000 Mbit/s (category 5e or 6UTP) 100 m Star. Full Duplex (switch)
1000BaseTX1000 Mbit/sBraided (category 6UTP)100 mStar. Full Duplex (switch)
1000BaseSX1000 Mbit/sOptical fibre (multimeter)550 mStar. Full Duplex (switch)
1000BaseLX1000 Mbit/sOptical fiber (monomode)5000 mStar. Full Duplex (switch)
10GBaseT 10000 Mbit/s (category 6a or 7UTP) 100 m
10GBaseLR 10000 Mbit/s Optical fiber (monomode) 10000 m
10GBaseSR 10000 Mbit/s Optical fibre (multimeter) 300 m

Commonly used hardware in an Ethernet network

The elements of an Ethernet network are: network card, repeaters, hubs, bridges, switches, network nodes and the interconnection medium. Network nodes can be classified into two large groups: data terminal equipment (DTE) and data communication equipment (DCE).

The DTEs are network devices that generate the destination of the data: PCs, workstations, file servers, print servers; they are all part of the group of final stations. The DCEs are the intermediary network devices that receive and retransmit the frames within the network; They can be: switches (switch), routers, concentrators (hub), repeaters or communication interfaces. For example: a modem or an interface card.

  • Network or NIC Interface Card, allows a computer to access a local network. Each card has one unique MAC address that identifies it in the network. A computer connected to a network is called No..
  • Repeater or repeater, increases the scope of a physical connection, receiving the signals and retransmitting them, to avoid their degradation, through the means of transmission, achieving a greater scope. It is usually used to unite two local areas 'of equal' technology and only has 'two' ports. Operates in the physical layer of the OSI model.
  • Concentrator or Hub, it works as a repeater but allows the interconnection of 'multiples' nodes. Its operation is relatively simple because it receives an ethernet plot, by one of its ports, and repeats it by all its remaining ports without running any process on them. Operates in the physical layer of the OSI model.
  • Network bridge or bridge, interconnects network segments making the change frames (tramas) between networks according to a board of addresses that tells you in which segment a given MAC address is located. They are designed for use between LANs that use identical protocols in the physical layer and MAC (media access control). Although they exist bridges more sophisticated that allow the conversion of different MAC formats (Ethernet-Token Ring For example).
Connections on an Ethernet switch.
  • Switch or switch switchIt works like bridge, but allows interconnection of multiple network segments, works at faster speeds and is more sophisticated. Them switches may have other features, such as virtual networks, and allow your settings through your own network. It works basically in the OSI layer 2 (data link). This is why they are able to process information from the plots; their most important functionality is in the steering tables. For example, a computer connected to port 1 of the switch sends a plot to another computer connected to port 2; the switch receives the plot and transmits it to all its ports, except the one where it was received; computer 2 will receive the message and will finally answer it, generating traffic in the opposite sense; now the switch You will know the MAC addresses of computers in port 1 and 2; when you receive another plot with destination direction from one of them, you will only transmit the plot to that port thus diminishing the traffic of the network and contributing to the good functioning of it.

Present and future of Ethernet

Ethernet was originally conceived as a protocol to meet the needs of local area networks (LANs).

Starting in 2001, Ethernet reached 10 Gbit/s which made the technology much more popular. Within the sector, ATM was considered as the total in charge of the upper levels of the network, but the 802.3ae standard (Gigabit Ethernet 10) has been in a good position to be extended to the WAN level.

Contenido relacionado

Reduced instruction set computing

In computer architecture, RISC is a type of CPU design generally used in microprocessors or microcontrollers with the following fundamental...

OSI model

The Open Systems Interconnection Model known as the OSI model, Open Ssystems Interconnection) is a reference model for network protocols created in 1980 by...

Imevisión

The Instituto Mexicano de la Televisión, also known as Imevisión, was a Mexican state agency in charge of operating television stations owned by the federal...
Más resultados...
Tamaño del texto:
undoredo
format_boldformat_italicformat_underlinedstrikethrough_ssuperscriptsubscriptlink
save