Local Area Networks Chapter 7 Ethernet LANs 635

Local Area Networks Chapter 7 Ethernet LANs 635. 412 Class #3 1

IEEE 802. 3 Family of LAN Protocols Ethernet n Introduction & History – – The most widely used LAN technology today is Ethernet and its successive generations – and it is to this day still evolving! Ethernet was originally developed in early 1973 at the Xerox Palo Alto Research Center (PARC); it originally ran at 2. 94 -Mbps In 19 xx Bob Metcalfe and Charlie Roll were granted Patent #4, 063, 220 Metcalfe persuaded Xerox to partner with other companies, resulting in the initial effort to standardize Ethernet n – The first widespread standard for 10 -Mbps Ethernet was ratified by DEC, Intel, and Xerox in 1980 with a major revision in 1983 The IEEE 802. 3 working group developed a vendor-neutral standard based on these earlier standards (though there are slight differences they are compatible) 635. 412 Spring 2005 Class #3: Ethernet LANs 2

IEEE 802. 3 Family of LAN Protocols Ethernet n What are the essential components of Ethernet? – There are three components of Ethernet to study n MAC Algorithm – n n Based on a mechanism called Carrier Sense Multiple Access with Collision Detect (CSMA/CD) MAC Frame Physical Layer Standards 635. 412 Spring 2005 Class #3: Ethernet LANs 3

IEEE 802. 3 Family of LAN Protocols Precursors to CSMA/CD and Ethernet n n CSMA/CD is a contention based mechanism that has evolved from several different earlier random access schemes The earliest contention mechanism used was called ALOHA was developed in the late 1960 s for use in packet radio networks at the University of Hawai’i How ALOHA works: – – n A station may transmit a frame at any time After a station transmits it listens to the medium for an amount of time equal to the maximum round-trip propagation time plus a small fixed increment If a station receives an acknowledgement during that time the frame was properly received, otherwise the transmitter resends the frame A received frame is checked for errors which may be due to noise or a collision (in either case the errored frame is ignored) While stations using ALOHA are very simple, it is very inefficient with a maximum utilization around 18% 635. 412 Spring 2005 Class #3: Ethernet LANs 4

Pure Aloha efficiency P(success by given node) = P(node transmits). P(no other node transmits in [p 0 -1, p 0] = p. (1 -p)N-1 = p. (1 -p)2(N-1) … choosing optimum p and then letting n -> infty. . . = 1/(2 e) =. 18 Even worse ! 635. 412 Spring 2005 Class #3: Ethernet LANs 5

IEEE 802. 3 Family of LAN Protocols Precursors to CSMA/CD and Ethernet n Later, an improvement known as slotted ALOHA was developed – – n The medium is divided into uniform time slots (this requires some kind of synchronization mechanism available to all stations) Transmission is permitted to begin only on slot boundaries; otherwise this works the same as regular ALOHA This forces collisions to overlap entirely, protecting transmissions somewhat from failure This increases efficiency to a maximum around 37% at the expense of a more complex system The core problem with both Aloha and slotted ALOHA was that they didn’t take advantage of the short propagation delay in the LAN/packet radio environment; better feedback about network conditions can be used to increase efficiency 635. 412 Spring 2005 Class #3: Ethernet LANs 6

Slotted ALOHA Pros n single active node can continuously transmit at full rate of channel n highly decentralized: only slots in nodes need to be in sync n simple 635. 412 Spring 2005 Cons n collisions, wasting slots n idle slots n nodes may be able to detect collision in less than time to transmit packet Class #3: Ethernet LANs 7

Slotted Aloha efficiency n Efficiency is the long-run fraction of successful slots when there’s many nodes, each with many frames to send n n n Suppose N nodes with many frames to send, each transmits in slot with probability p prob that 1 st node has success in a slot For max efficiency with N nodes, find p* that maximizes Np(1 -p)N-1 For many nodes, take limit of Np*(1 -p*)N-1 as N goes to infinity, gives 1/e =. 37 = p(1 -p)N-1 n prob that any node has a success = Np(1 -p)N-1 635. 412 Spring 2005 Class #3: Ethernet LANs At best: channel used for useful transmissions 37% of time! 8

IEEE 802. 3 Family of LAN Protocols Precursors to CSMA/CD and Ethernet n Carrier Sense Multiple Access (CSMA) – – – Experience with ALOHA and slotted ALOHA led to the development of a new technique called Carrier Sense Multiple Access (CSMA) With CSMA a station wishing to transmit first listens to the medium to see if it in use; if so it waits until the medium is idle If a collision occurs or no acknowledgement is received the station must retransmit the frame CSMA works best in a medium where the propagation delay is small and stations quickly determine that someone else is transmitting The maximum utilization with CSMA is far better than ALOHA or slotted ALOHA, but varies with the frame size and propagation time 635. 412 Spring 2005 Class #3: Ethernet LANs 9

CSMA collisions spatial layout of nodes collisions can still occur: propagation delay means two nodes may not hear each other’s transmission collision: entire packet transmission time wasted note: role of distance & propagation delay in determining collision probability 635. 412 Spring 2005 Class #3: Ethernet LANs 10

IEEE 802. 3 Family of LAN Protocols Carrier Sense Multiple Access (CSMA) n Persistance – – If a station wishing to transmit finds the medium idle, there are several different methods for how the station should react With nonpersistant CSMA, if the medium is idle the station waits a random amount of time before transmitting n n n Random delay helps reduce the possibility of collisions Utilization of the medium is affected by the delay because the medium will generally remain idle even though one or more stations have frames to transmit In effect, the station acts very conservatively when competing for the medium 635. 412 Spring 2005 Class #3: Ethernet LANs 11

IEEE 802. 3 Family of LAN Protocols Carrier Sense Multiple Access (CSMA) n Persistance – To increase efficiency 1 -persistent CSMA can be used n n n In this case once the medium is found to be idle a station transmits immediately Very efficient at low loads however, if two or more stations are waiting to transmit they will collide (which could induce gridlock) In effect this makes stations very greedy, automatically trying to capture the medium 635. 412 Spring 2005 Class #3: Ethernet LANs 12

IEEE 802. 3 Family of LAN Protocols Carrier Sense Multiple Access (CSMA) n Persistance – A compromise to these two extremes is to use ppersistent CSMA n n In this case if the medium is found idle the station transmits immediately with probability p or otherwise delays transmission for one time unit If the station delays it will again have to sense that the medium is idle before transmitting The choice of p is very important; too high can cause too many collisions if a lot of stations wish to transmit while having p too low will result in long delays for stations with frames to send An adaptive value for p is an active area of LAN research 635. 412 Spring 2005 Class #3: Ethernet LANs 13

IEEE 802. 3 Family of LAN Protocols CSMA with Collision Detection (CSMA/CD) n More enhancements to CSMA – – While CSMA is a big improvement over ALOHA, it still has one key efficiency problem: dealing with collisions This can be quite a blow to efficiency if a station continues to transmit the whole frame even though it has collided with another transmission (especially for long frames) The fix is to ensure that the transmitting station continues to listen while transmitting and if a collision is detected stop transmitting as soon as possible The specific CSMA/CD procedure: a) If medium is idle then transmit, if not go to (b) b) If medium is busy, listen for medium to become idle & transmit c) If a collision is detected during transmission transmit a brief jamming signal and stop d) Wait a random amount of time and return to (a) 635. 412 Spring 2005 Class #3: Ethernet LANs 14

CSMA/CD collision detection 635. 412 Spring 2005 Class #3: Ethernet LANs 15

IEEE 802. 3 Family of LAN Protocols CSMA with Collision Detection (CSMA/CD) n Description of CSMA/CD – Example of CSMA/CD operation [Figure 7. 2] n n – – Signal Propagation Signal Overlap Collision Detection by the first station Collision Detection by the second station With CSMA/CD the wasted utilization on the medium is reduced to the amount of time it takes to detect a collision (twice the maximum propagation time of the network) In order to ensure collision detection all frames must be long enough to allow collision detection before the end of transmission (or it wouldn’t be collision detection!) 635. 412 Spring 2005 Class #3: Ethernet LANs 16

IEEE 802. 3 Family of LAN Protocols CSMA/CD Operation 635. 412 Spring 2005 Class #3: Ethernet LANs 17

IEEE 802. 3 Family of LAN Protocols CSMA/CD and Persistent Retransmission n Though the basic 1 -persisent protocol can have performance problems, an enhanced version is used with Ethernet – n The specific method used in Ethernet is called binary exponential backoff – – n This is possible because the time wasted with collisions is usually short & all stations use a random backoff after a collision (so someone retransmits quickly) Stations try to transmit repeatedly after collisions, but each time the mean delay is doubled (delay(k)=random[0, (2^k -1)x slot time]) After 16 unsuccessful tries, the station gives up and reports an error With a 1 -persistent protocol and binary exponential backoff Ethernet is efficient over a wide range of loads – – Minimizes transmission delay at low loads while sorting out collisions quickly at higher loads Unfortunately this combination of techniques has a LIFO effect with stations wishing to transmit 635. 412 Spring 2005 Class #3: Ethernet LANs 18

IEEE 802. 3 Family of LAN Protocols Carrier Sense and Collision Detection n So what really is a Collision? – – With baseband Ethernet systems the ‘carrier’ is sensed by detecting the presence of voltage transitions on the medium A collision is detected if a station detects a signal exceeding the maximum that could be transmitted by the station alone n n n – Because of attenuation this may not work if stations far apart suffer a collision This limits the maximum span of the network; in the IEEE 802. 3 standard the limit is 500 m Since repeaters are supposed to allow transparent expansion of the network, they must pass collisions In star-wired twisted pair networks collisions are handled in a much simpler way, since collision detection is done with logic circuits instead of voltage sensing 635. 412 Spring 2005 Class #3: Ethernet LANs 19

IEEE 802. 3 Family of LAN Protocols Carrier Sense and Collision Detection n Operation in Star-wired twisted pair networks – – For any hub, if frames are received simultaneously on two or more inputs a collision has occurred A special collision presence signal is generated at all hub ports and is propagated throughout a hierarchal star topology [Figure 7. 3] 635. 412 Spring 2005 Class #3: Ethernet LANs 20

635. 412 Spring 2005 Class #3: Ethernet LANs 21

IEEE 802. 3 Family of LAN Protocols Summary of CSMA/CD Operation 1. 2. 3. If the medium is idle for at least the IFG (inter-frame gap) time interval (96 bit times), then transmit the frame immediately If the medium is busy, then wait until the medium is idle and start with step #1 If a collision is detected during transmission, continue to send data until the complete preamble and 32 bits of data are sent (the ‘jamming’ signal) – – – The jam signal should be long enough to ensure all other stations are alerted to a collision The stations involved in the collision wait a random (backoff) time and start at step #1 These station keep a collision counter to track the number of collisions that have occurred in the process of transmitting a specific frame 635. 412 Spring 2005 Class #3: Ethernet LANs 22

IEEE 802. 3 Family of LAN Protocols Summary of CSMA/CD Operation – 4. Once a station has transmitted the preamble and 512 bits of a frame (called the slot time), the station has acquired the medium – – 5. If a collision occurs and the collision counter is greater than zero, the random backoff time is doubled After this point all other stations should be aware of the transmission, so no other station should attempt to transmit Late collisions are serious errors; a typical cause is a duplex mismatch on a link Once a frame is successfully transmitted the collision counter is cleared so the backoff process starts anew 635. 412 Spring 2005 Class #3: Ethernet LANs 23

IEEE 802. 3 Family of LAN Protocols Ethernet at the MAC Layer n MAC frame format for the IEEE 802. 3 protocol 635. 412 Spring 2005 Class #3: Ethernet LANs 24

IEEE 802. 3 Family of LAN Protocols Ethernet at the MAC layer n MAC frame format for the IEEE 802. 3 protocol – – – – Preamble: 56 bits of alternating ones & zeros to allow bit synch Start Frame Delimiter: the bit sequence 10101011 Destination Address (DA): 48 bit unicast, broadcast, or multicast address Pad: necessary to ensure short frames allow collision detection to occur Source Address (SA): 48 bit unicast address Length/Type: Field contains either the actual length of the frame (up to a maximum of 1518 bytes) if a IEEE 802. 3 standard frame or the payload type if frame conforms to earlier Ethernet standard LLC Data: the LLC PDU Frame Check sequence (FCS): a 32 bit cyclic redundancy check 635. 412 Spring 2005 Class #3: Ethernet LANs 25

Exercises n Question 1: Find the Medium length when Slot time (ST) and data rate (R) are (standards for 802. 3): – – – n ST=512 bits and R=10 Mbps, ST=512 bits and R=100 Mbps, ST=4096 bits and R=100 Mbps, Question 2: The header of IP, TCP, LLC, and MAC are 20 bytes, 2 bytes, and 26 bytes, respectively. Assume a frame F contains a payload of 64 bytes (64 -1518 bytes for 802. 3). What should be the MAC frame padding size to transmit F on a LAN where Tp=51. 2 micro seconds and R=100 Mbps. 635. 412 Spring 2005 Class #3: Ethernet LANs 26

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n Introduction to the Physical Layer – – The IEEE 802. 3 committee has been very good at responding to the evolution of technology by introducing new physical layer media, but that has led to enough options to make the choice daunting for customers Physical Layer Medium Alternatives [Table 7. 2] n n n 1 BASE-T 10 BASE-5 10 BASE-2 10 BASE-T 10 BROAD-36 10 BASE-F 635. 412 Spring 2005 Class #3: Ethernet LANs 27

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n Medium Attachment Unit – – – The IEEE 802. 3 standard provides a solution for the situation where the networked station is some distance from the actual attachment to the Ethernet transmission medium This requires the transmission functionality to be split into two components The piece of equipment actually connected to the medium is called the Medium Attachment Unit (MAU) and performs the following functions: n n n Transmit to & receive signals from the medium Recognize the presence of signals on the medium Recognize a collision 635. 412 Spring 2005 Class #3: Ethernet LANs 28

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n Medium Attachment Unit – – The MAU is connected to the rest of the network interface hardware at the station by a set of cables called the Attachment Unit Interface (AUI) that meet the relevant IEEE 802. 3 specifications AUI details n n n Detailed electrical interface specifications Uses D-type 9 or (more common) 15 pin connector 8 shielded wires typically used: one pair for Transmit and one pair for Receive Data One pair for Collision signaling Power is also supplied through interface to MAU Maximum distance of 50 m 635. 412 Spring 2005 Class #3: Ethernet LANs 29

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n Medium Attachment Unit 635. 412 Spring 2005 Class #3: Ethernet LANs 30

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n 10 BASE-5 Medium Specification – – The original 802. 3 physical medium specification based directly on the earlier DIX Ethernet specification Special purpose 50 ohm coax cable & baseband Manchester signaling is used The maximum length of any coaxial segment is 500 m, a maximum of four repeaters between any two stations allows a maximum network size of 2. 5 km There can only be one path between any two stations; loops are not allowed 635. 412 Spring 2005 Class #3: Ethernet LANs 31

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n 10 BASE-2 Medium Specification – – This physical layer specification was developed to provide a cheaper, easier to install alternative to 10 BASE-5 for personal computer networking Thinner 50 ohm coaxial cable is used and all electronics are housed on the Network Interface card (NIC) Higher attenuation and noise figures for the thinner cable limits the maximum cable size and number of stations It is possible to interconnect 10 BASE-5 and 10 BASE-2 segments using appropriate repeaters, but a 10 BASE 2 segment should not be used to interconnect two 10 BASE-5 segments 635. 412 Spring 2005 Class #3: Ethernet LANs 32

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n 10 BASE-F Medium Specification – – – Even though fiber was very expensive and unusual to use at the time, the IEEE considered it important to provide a fiber optic transport option All variants use a pair of multi-mode fiber: one TX & one RX fiber (ST connectors) Physical signaling is intensity modulation with Manchester encoding The original option was called FOIRL (Fiber Optic Inter. Repeater Link) Later, FOIRL was replaced by a more comprehensive specification with three options for fiber-based Ethernet n n n 10 BASE-FL 10 BASE-FB 10 BASE-FP 635. 412 Spring 2005 Class #3: Ethernet LANs 33

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n FOIRL and 10 BASE-FL – The FOIRL standard was first published officially in 1989 n n – Designed to allow easy extension of an Ethernet segment between two buildings (point-to-point) Maximum distance of 1000 meters Only for repeater <-> repeater links Some vendors developed FOIRL capable stations but this functionality never standardized Replaced in 1993 with an enhanced standard called 10 BASE-FL n n n Backwards compatible with FOIRL repeaters Maximum distance of 2000 meters Can connect together two repeaters, two stations, or a station and a repeater 635. 412 Spring 2005 Class #3: Ethernet LANs 34

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n 10 BASE-FB and 10 BASE-FP – – Two other variants standardized at the same time as 10 BASE-FL for other uses 10 BASE-FB defines synchronous signaling fiber backbone segment n n n – Allows multiple repeaters to be linked over the normal repeater limits Used to build lengthy backbone segments Not currently used in commercially available products 10 BASE-FP defines a fiber-based Ethernet segment that can have up to 33 stations n n Uses a passive fiber equipment & couplers to build a network up to 500 meters in diameter Never developed into commercially available equipment 635. 412 Spring 2005 Class #3: Ethernet LANs 35

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n 10 BASE-T Medium Specification – – By using a star topology and sacrificing some distance on the length of cable runs twisted pair can be used as a 802. 3 physical layer medium All stations are connected in a physical star topology back to a central hub/repeater with two twisted pairs (one send/one receive) The length of each link is limited to 100 m; like the other physical layers specifications there is a maximum span of five ‘segments’ and four repeaters It is possible to mix 10 BASE-T with 10 BASE-5 and 10 BASE-2 segments (the maximum number of coax segments in a ‘mixed’ network is three) 635. 412 Spring 2005 Class #3: Ethernet LANs 36

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n 10 BASE-T Physical Layer – – – Requires two pair of Category 3 UTP (or better) Uses Manchester signaling across a differential interface The familiar RJ-45 interface is used with the following pin configuration: n n n n Pin Pin 635. 412 Spring 2005 1: 2: 3: 4: 5: 6: 7: 8: TD+ TDRD+ Unused RDUnused Class #3: Ethernet LANs 37

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n 10 BASE-T Hubs – Hubs have the following functionality and do not care whether a port is connected to a station or another hub: n n n – A valid signal on any port is repeated on all other links If a collision occurs a collision presence signal is generated on all ports If a collision presence signal is detected on a port, it is repeated on all other links More sophisticated hubs will have some network management capability n n Collection of statistics Disconnecting ‘misbehaving’ ports 635. 412 Spring 2005 Class #3: Ethernet LANs 38

IEEE 802. 3 Family of LAN Protocols 10 -Mbps Ethernet n Simple 10 BASE-T configuration with Hubs [Fig. 7. 5] 635. 412 Spring 2005 Class #3: Ethernet LANs 39

IEEE 802. 3 Family of LAN Protocols 100 -Mbps Ethernet n Introduction – – Also known as Fast Ethernet, this technology was first developed by Kalpana in 1991 -1992 Shortly afterward, a standard specification was developed by the IEEE 802. 3 u subcommittee to fill the need for a low -cost, Ethernet-compatible 100 -Mbps LAN technology All of the 100 BASE-T physical layer options use the standard IEEE 802. 3 MAC protocol and frame format Like 10 -Mbps Ethernet a number of physical layer options were developed to allow network designers more flexibility n n Options were developed for fiber optic and twisted pair cabling In addition, a standard interface between the MAC and physical (PHY) layers was developed to facilitate this 635. 412 Spring 2005 Class #3: Ethernet LANs 40

IEEE 802. 3 Family of LAN Protocols 100 -Mbps Ethernet 100 BASE-T Options n 100 BASE-X (2 links: Transmit & Receive): – 100 BASE-TX uses TP and UTP (2 CAT 5 UTP or 2 STP) and MLT-3. – 100 BASE-FX uses 2 FO (T & R) using 4 B/5 B NRZI n 100 BASE-T 4 uses 4 CAT 3 or CAT 5 UTP and 8 B 6 T 635. 412 Spring 2005 Class #3: Ethernet LANs 41

IEEE 802. 3 Family of LAN Protocols 100 -Mbps Ethernet Physical Layer n 100 BASE-X Physical Layer Specifications – – The two transmission options under the 100 BASE-X specification provide a full duplex 100 -Mbps link operating over a pair of physical links (one send/one receive) The 100 BASE-TX option uses two pairs of twisted pair cabling for the connection n n – Either Category 5 or Shielded twisted pair can be used with a maximum segment length of 100 m A signaling scheme called MLT-3 is used A later standard called 100 BASE-T 2 allowed fast Ethernet to run over two pairs of Category 3 cable; however it has never been implemented 635. 412 Spring 2005 Class #3: Ethernet LANs 42

MLT-3 Signaling 635. 412 Spring 2005 Class #3: Ethernet LANs 43

Digital Signal Encoding for 100 BASE-T n MLT-3 Encoding – – – While 4 B/5 B is great for fiber optic links, it’s spectral characteristics are unsuitable for UTP (radiates a large amount of RF interference) Used with 100 BASE-TX and the twisted pair version of FDDI A ternary voltage encoding scheme that produces a transition for every binary one The 4 B/5 B and NRZ-I encoding schemes are still used, but after that the elements are reconverted to NRZ, scrambled, and then encoded using MLT-3 for transmission Example MLT-3 encoding [Figure 7. 16]: 635. 412 Spring 2005 Class #3: Ethernet LANs 44

IEEE 802. 3 Family of LAN Protocols 100 -Mbps Ethernet n 100 BASE-X Physical Layer Specifications – The 100 BASE-FX option uses one pair of fiber optic cabling and a 4 B/5 B-NRZI encoding scheme n n n On fiber a binary signaling mechanism (intensity modulation) is used Essentially the ANSI X 3 T 9. 5 FDDI physical layer standards have been reused The standards address operation over Multimode Fiber cable (up to 2 km); there are ‘proprietary’ 100 BASE-FX products that operate over Single Mode Fiber (up to 10 km) 635. 412 Spring 2005 Class #3: Ethernet LANs 45

IEEE 802. 3 Family of LAN Protocols 100 -Mbps Ethernet Physical Layer Options n 100 BASE-T 4 – – Another transmission option which allows the use of lower grade Category 3 cabling for data transmission Four pairs are needed for 100 BASE-T 4; each direction of the 100 Mbps data link is split equally over three pairs n n – Pairs 1, 3, and 4 are used for data transmission while pairs 2, 3, & 4 are used for data reception Pairs 3 and 4 must be configured for bi-directional communication Pair 2 is also used for collision detection Note that true full-duplex [symmetrical speeds] operation is not possible! To provide high data rates over Category 3 UTP a ternary signaling scheme called 8 B 6 T is used 635. 412 Spring 2005 Class #3: Ethernet LANs 46

IEEE 802. 3 Family of LAN Protocols Digital Signal Encoding for 100 BASE-T n 8 B 6 T Encoding – Another ternary encoding scheme used with 100 BASE -T 4 where a block of 8 data bits is encoded into a block of 6 ternary symbols n n n – +V, 0, and –V are three voltage levels used The resulting stream is transmitted round-robin across the three transmit pairs leading to a signaling rate of 25 -Mbaud on each pair The mapping table [Table 7. 6] outlines the 8 B/6 T code translation; the mappings were chosen to maintain proper timing and dc balance (under certain circumstances 6 T code groups can be inverted for transmission to help maintain DC balance) Theoretically similar the 4 B/5 B and other block encoding schemes except ternary symbols are used 635. 412 Spring 2005 Class #3: Ethernet LANs 47

100 BASE-T 4 Category 3 UTP and 8 B 6 T signaling How to get 100 Mbps using 3 UTPs (fourth UTP for collision) • 8 binary (256 levels) encoded as 6 Ternary, • 6 Ternary (729 levels): to each combination of 8 -bit we have one combination (transmittable) of 6 ternary signals, • Each UTP is assigned 2 ternary (3 UTP get 6 ternary), • Since each UTP (25 Mbaud) gets 2 T, then the data rate is (25/2) x 8 bits = 100 Mbps. 635. 412 Spring 2005 Class #3: Ethernet LANs 48

IEEE 802. 3 Family of LAN Protocols 100 -Mbps Ethernet n Another variant: 100 VG-Any. LAN – – – An alternative to the CSMA-CD protocols that was developed by the IEEE 802. 12 task force Uses a MAC technique called demand priority (round robin with priority enhancements) Uses all four pairs with 5 B 6 B encoding in half-duplex operation Supports the 802. 3 MAC frame format for interoperability with regular Ethernet Originally a competitor to 100 BASE-T – never been widely adopted 635. 412 Spring 2005 Class #3: Ethernet LANs 49

IEEE 802. 3 Family of LAN Protocols 100 -Mbps Ethernet n 100 BASE-X Media Independent Interface (MII) – – This was developed with the standard to allow different but interoperable physical layers to be developed for a single MAC layer Defines an electrical and mechanical interface for Fast Ethernet similar in function to the 10 -Mbps AUI n n n – Uses 40 -pin miniature D-type connector Uses digital-logic grade signals capable of a 0. 5 m maximum distance Separate clock, data, and power pins; 4 bits transferred in parallel on both transmit and receive The MII was rarely used; most 100 BASE-X equipment has the transceiver built on the NIC 635. 412 Spring 2005 Class #3: Ethernet LANs 50

IEEE 802. 3 Family of LAN Protocols 100 BASE-T Ethernet n Configuration and Operation – – – A 100 BASE-T network is a physical star and logical CSMA-CD bus; the central hub/repeater performs the same functions as the equivalent 10 BASE-T component A single CSMA/CD network is sometimes referred to as a collision domain; bridges and routers separate collision domains while repeaters are part of a single collision domain The 100 BASE-T standard defines two types (grades) of repeaters: Class I and Class II 635. 412 Spring 2005 Class #3: Ethernet LANs 51

IEEE 802. 3 Family of LAN Protocols 100 BASE-T Ethernet n Configuration and Operation – Details on Class I and II repeaters n n n Class I repeaters support connection of different types of physical media segments; only a single Class I repeater can be used in a collision domain Class II repeaters are limited to a single physical media type; two Class II repeaters can be used in a collision domain Like 10 -Mbps Ethernet the connection of bridges and other hub/repeaters to a hub/repeater port is treated the same as a station connected to the hub/repeater 635. 412 Spring 2005 Class #3: Ethernet LANs 52

IEEE 802. 3 Family of LAN Protocols 100 BASE-T Ethernet n A Typical Mixed Ethernet Environment 635. 412 Spring 2005 Class #3: Ethernet LANs 53

IEEE 802. 3 Family of LAN Protocols 100 BASE-T Ethernet Switches n n n Essentially a multi-interface bridge layer 2 (frame) forwarding, filtering using LAN addresses Switching: A-to-A’ and B-to-B’ simultaneously, no collisions large number of interfaces often: individual hosts, starconnected into switch Ethernet, but no collisions! cut-through switching: frame forwarded from input to output port without awaiting for assembly of entire frame – slight reduction in latency combinations of shared/dedicated, 10/1000 Mbps interfaces – n n 635. 412 Spring 2005 Class #3: Ethernet LANs 54

IEEE 802. 3 Family of LAN Protocols 100 -Mbps Ethernet Autonegotiation Option n Introduction – – An optional feature of the IEEE 802. 3 100 BASE-T standards which defines a method for two devices connected to the same link to negotiate physical and MAC layer options Autonegotiation takes place using a data exchange ‘buried’ in the link integrity pulses exchanged during initialization & idle periods on the link n n n Normal Link Integrity pulses that take place once every 210 sec are replaced by a series of fast link pulses (FLP) 10 BASE-T and 100 BASE-T devices that do not support autonegotiation are supposed to ignore the data exchange Autonegotiation does not in any way interfere with data exchange 635. 412 Spring 2005 Class #3: Ethernet LANs 55

IEEE 802. 3 Family of LAN Protocols Autonegotiation n n The autonegotiation sequence consists of 33 pulses; the 17 odd pulses form the normal link integrity sequence, the 16 even pulses form the autonegotiation code word The 16 bit code word contains the following fields: – – Selector field (5 bits): identifies the message type Technology ability field (8 bits): specifies what different medium technologies the device can support – five are now defined (from most to least preferred): n n n 100 BASE-TX full-duplex 100 BASE-T 4 100 BASE-TX 10 BASE-T full-duplex 10 BASE-T The most preferred technology supported by both devices will be used 635. 412 Spring 2005 Class #3: Ethernet LANs 56

IEEE 802. 3 Family of LAN Protocols Autonegotiation n The 16 bit autonegotiation code word contains the following fields: – – – n Remote Fault bit: if set informs the remote device that a link fault has occurred Acknowledge bit: indicates that the remote device on the link successfully received the autonegotiation code word Next page bit: indicates that another code word with more specific information will follow (e. g. - error specification) The original standard has also been extended to higher speed Ethernet variants (so there would be 10/1000 BASE-T interfaces) 635. 412 Spring 2005 Class #3: Ethernet LANs 57

IEEE 802. 3 Family of LAN Protocols Gigabit Ethernet n Introduction – – – Knowing development of a 100 -Mbps Ethernet standard would only temporarily satisfy users needs for faster LAN technologies, in 1995 the IEEE 802. 3 committee began work on Gigabit Ethernet The final 802. 3 z standard was approved in June, 1998 though work on a twisted pair PHY took longer IEEE 802. 3 z highlights n n n Gigabit Ethernet retains the 802. 3 MAC frame format, so it is interoperable with 10 -Mbps and 100 -Mbps Ethernets A number of physical layer options were defined It also retains half and full-duplex modes, but some modifications were needed to allow a reasonable network size with CSMA/CD 635. 412 Spring 2005 Class #3: Ethernet LANs 58

Gigabit Ethernet Table 6. 3 IEEE 802. 3 1 Gbps Fast Ethernet medium alternatives Medium Max. Segment Length Topology n n n 1000 base. SX 1000 base. LX 1000 base. CX 1000 base. T Optical fiber multimode Two strands Optical fiber single mode Two strands Shielded copper cable Twisted pair category 5 UTP 550 m 5 km 25 m 100 m Star Slot time increased to 512 bytes (4096 bits) Small frames need to be extended to 512 B Frame bursting to allow stations to transmit burst of short frames Frame structure preserved but CSMA-CD essentially abandoned Extensive deployment in backbone of enterprise data networks and in server farms 635. 412 Spring 2005 Class #3: Ethernet LANs 59

IEEE 802. 3 Family of LAN Protocols Gigabit Ethernet 635. 412 Spring 2005 Class #3: Ethernet LANs 60

IEEE 802. 3 Family of LAN Protocols Gigabit Ethernet n Typical Application Scenario 635. 412 Spring 2005 Class #3: Ethernet LANs 61

Typical Ethernet Deployment Server farm Server Switch/router Server Ethernet switch 100 Mbps links Hub 10 Mbps links Department A 635. 412 Spring 2005 Server Gigabit Ethernet links Ethernet switch 100 Mbps links Server Hub 10 Mbps links Department B Class #3: Ethernet LANs Switch/router Ethernet switch 100 Mbps links Server Hub 10 Mbps links Department C 62

IEEE 802. 3 Family of LAN Protocols Gigabit Ethernet n Protocol Architecture 635. 412 Spring 2005 Class #3: Ethernet LANs 63

IEEE 802. 3 Family of LAN Protocols Gigabit Ethernet n Protocol Architecture – Like earlier technologies Gigabit Ethernet allows a choice of physical media n – A 8 B/10 B encoding scheme is used with fiber optic and coax cabling while a very complex scheme called 4 DPAM 5 is used with UTP A special interface called the Gigabit media-dependent interface (GMII) has been defined that lets users mixand-match MAC & Physical layer parts n n n GMII uses dual synchronous 8 bit buses for low-level chipto-chip interfacing Unlike the 100 BASE-X MII, does not specify a physical connector Has allowed the development of GBICs (Gigabit Interface Converter) – pluggable Physical Media modules for Gigabit Ethernet equipment 635. 412 Spring 2005 Class #3: Ethernet LANs 64

8 B/10 B encoding scheme 635. 412 Spring 2005 Class #3: Ethernet LANs 65

IEEE 802. 3 Family of LAN Protocols Gigabit Ethernet n Example GBIC Module 635. 412 Spring 2005 Class #3: Ethernet LANs 66

IEEE 802. 3 Family of LAN Protocols Gigabit Ethernet MAC n Media Access Layer – Gigabit Ethernet is used for full-duplex operation, but it is possible to operate at half-duplex. The IEEE added two enhancements to CSMA/CD so it would work in a highspeed environment n n – Carrier Extension: a variable set of special symbols is appended to short frames to make them a minimum of 4096 bits long (this is to keep the frame transmission time longer than the propagation time) Frame Extension: allows multiple small frames to be transmitted consecutively without relinquishing control of the media The Gigabit Ethernet specification also enhances the flow control specification to allow asynchronous flow control n n 635. 412 Spring 2005 Flow control can be used in one direction but not the other Can be negotiated via the autonegotiation feature Class #3: Ethernet LANs 67

IEEE 802. 3 Family of LAN Protocols Gigabit Ethernet PHY n Physical Layer Choices – There are currently four alternatives defined by the IEEE 802. 3 committee: n n 635. 412 Spring 2005 1000 BASE-LX: specifies a long-reach alternative that supports fiber optic links of up to 5 km on single mode cable (550 m on multimode cable) 1000 BASE-SX: specifies a less expensive short-reach option that supports fiber optic links of up to 550 m on multimode fiber 1000 BASE-CX: supports short (25 m) links using a pair of special shielded twisted pair cables (one transmit/one receive) and encoding borrowed from the Fibre Channel PHY. 1000 BASE-T: uses four pairs of Category 5 unshielded twisted pair to support links up to 100 m (this particular option is known as 802. 3 ab) Class #3: Ethernet LANs 68

IEEE 802. 3 Family of LAN Protocols The Future of Ethernet n 10 -Gbps Ethernet – – The rapid growth in use of Gigabit Ethernet has spurred the need for even higher speed network technologies Specification was completed in 2002 for the definition of 10 -Gbps Ethernet designed for use in a variety of highspeed applications: n n – – Metropolitan Area Networks LAN Backbones Storage-Area Networks Long-haul carrier networks Work is being done by the IEEE 802. 3 ae task force Unlike previous versions, no half-duplex (CSMA/CD) option will be specified; full-duplex operation allows almost unlimited reach 635. 412 Spring 2005 Class #3: Ethernet LANs 69

IEEE 802. 3 Family of LAN Protocols 10 Gigabit Ethernet n Other key aspects of the IEEE 802. 3 ae standard – Will support the Ethernet MAC frame and other parameters defined in earlier standards (minimum and maximum frame sizes, etc. ) – Will define two Physical (PHY) layer interfaces distinguished by different encoding schemes: n n – The LAN interface will be the same as Gigabit Ethernet, only faster The WAN interface includes extra functionality that will frame the 10 Gb. E MAC frame within a simplified SONET OC-192 frame to allow direct interfacing to SONET multiplexers PHY layer standards defined for star-wired Multimode and Single mode fiber optic installations 635. 412 Spring 2005 Class #3: Ethernet LANs 70

IEEE 802. 3 Family of LAN Protocols 10 Gigabit Ethernet n Protocol Architecture Media Access Control (MAC) Full Duplex 10 Gigabit Media Independent Interface (XGMII) 10 Gigabit Attachment Unit Interface WWDM LAN PHY (8 B/10 B) WWDM PMD 1310 NM Serial LAN PHY (64 B/66 B) Serial PMD 850 nm Serial PMD 1310 nm Serial WAN PHY (64 B/66 B + WIS) Serial PMD 1550 nm Serial PMD 850 nm Serial PMD 1310 nm Serial PMD 1550 nm The Architectural Components of the 802. 3 ae Standard (WWDM: Low cost Wide Wavelength Division multiplexing, PMD: Phys. Media dependent interface) 635. 412 Spring 2005 Class #3: Ethernet LANs 71

IEEE 802. 3 Family of LAN Protocols Digital Signal Encoding for Gigabit Ethernet n 8 B/10 B Encoding – – An encoding scheme similar to 4 B/5 B that was developed by IBM for an earlier mainframe transmission system The developers designed the encoding scheme to provide the following advantages: n n – – Implemented with simple and reliable components Well balanced, which eliminates the need for scrambling Provide sufficient signaling transitions to ensure good synch Provides good error detection capability Though the actual coding is broken into a combination of a 5 B/6 B and a 3 B/4 B encoding scheme, this wasn’t absolutely necessary A process called disparity control can invert code words before transmission to keep the bit stream ‘balanced’ (an equal number of ones and zeros in the symbol stream) 635. 412 Spring 2005 Class #3: Ethernet LANs 72

IEEE 802. 3 Family of LAN Protocols Digital Signal Encoding for Gigabit Ethernet n 4 D-PAM 5 Encoding – – A complex scheme used for 1000 BASE-T over twisted pair; required to provide a required data rate with an acceptable level of bit errors The 4 D-PAM 5 has the following key features: n Uses four pair of Category 5 cable; each cable operates in full-duplex mode and carries 250 -Mbps in each direction n PAM 5 (Pulse Amplitude Modulation) is used to send two bits per symbol; keeps the signaling rate below 125 Mbaud n Uses Forward Error Correction to combat crosstalk and noise n Scrambling required to randomize signaling stream before transmission 635. 412 Spring 2005 Class #3: Ethernet LANs 73

IEEE 802. 3 Family of LAN Protocols Scrambling n Data Transmission and Scrambling – – – A technique for randomizing data because long strings of ones or zeros can cause problems in transmission systems (e. g. – loss of timing) A scrambler consists of a set of feedback registers & exclusive OR operations Different scrambling functions can be used based on the amount of scrambling desired & the complexity of the system n – 100 BASE-TX uses Bm = Am X 9 X 11 At the receiver the same scrambling algorithm is used on the received (scrambled) data stream to recover the ‘real’ data stream 635. 412 Spring 2005 Class #3: Ethernet LANs 74

IEEE 802. 3 Family of LAN Protocols Scrambling n Binary Logic Diagram of a Typical Scrambler Circuit 635. 412 Spring 2005 Class #3: Ethernet LANs 75

IEEE 802. 3 Family of LAN Protocols 10 Gigabit Ethernet n MAC-PHY Interface – – Although the basic MAC sublayer did not change, the interface to the Physical Layer required modifications Basic Interface: XGMII n n n – 74 conductor high-speed parallel interface based on 100 BASE-T MII (32 data paths each way) Separate clocking circuits Maximum length of interface is 6 cm Enhanced Interface: XAUI n n Concept came from the original AUI standard Uses four bi-directional differential serial data paths, each at 2. 5 Gbps Has a much lower pin count and is self-clocking Maximum practical distance of 50 cm 635. 412 Spring 2005 Class #3: Ethernet LANs 76

IEEE 802. 3 Family of LAN Protocols 10 Gigabit Ethernet n The Physical Layer – Consists of two sublayers n n 635. 412 Spring 2005 Physical Coding Sublayer (PCS) – Performs line coding for interface – Serializes and multiplexes bit stream – On the WAN PHY interfaces a special sublayer called the WIS (WAN interface sublayer) provides the functionality (framing, etc) required to interface to SONET networks Physical Medium Dependent (PMD) – Provides the physical interface to the transmission medium (e. g. – a laser for fiber cable) – Allows flexibility in deployment by allowing a number of different media options Class #3: Ethernet LANs 77

IEEE 802. 3 Family of LAN Protocols 10 Gigabit Ethernet n LAN PHY Options – Serial PCS LAN PHY Options n n – Uses 64 B/66 B line coding 10 GBASE-SR: 850 nm optical interface for short-haul links up to 300 meters on multi-mode fiber 10 GBASE-LR: 1310 nm optical interface for links up to 10 km on single-mode fiber 10 GBASE-ER: 1550 nm optical interface for long distance links up to 40 km CWDM PCS LAN PHY Option n n 635. 412 Spring 2005 Uses 8 B/10 B line coding using four 3. 125 Gbps wavelengths on a single fiber (Coarse Wavelength Division Multiplexing) 10 GBASE-L 4: 1310 nm optical interface for use on multimode (300 m) or single-mode fiber (10 km) Class #3: Ethernet LANs 78

IEEE 802. 3 Family of LAN Protocols 10 Gigabit Ethernet n WAN PHY Options – – – WIS allows the WAN PHY options to run over a SONET OC-192 link or wavelength (nominal line rate of 9. 953 Gbps) These interfaces will appeal to service providers and carriers with SONET-based networks Serial PCS WAN PHY Options n n 635. 412 Spring 2005 Uses 64 B/66 B line coding 10 GBASE-SW: 850 nm optical interface for short-haul links up to 300 meters on multi-mode fiber 10 GBASE-LW: 1310 nm optical interface for links up to 10 km on single-mode fiber 10 GBASE-EW: 1550 nm optical interface for long distance links up to 40 km Class #3: Ethernet LANs 79

IEEE 802. 3 Family of LAN Protocols 10 Gigabit Ethernet Table 6. 5 IEEE 802. 3 10 Gbps Ethernet medium alternatives 10 Gbase. SR Medium Two optical fibers Multimode at 850 nm 64 B 66 B code Max. Segment Length n n n 300 m 10 GBase. LR 10 Gbase. EW Two optical fibers Single-mode at 1310 nm Single-mode at 1550 nm SONET compatibility 64 B 66 B 10 km 40 km 10 Gbase. LX 4 Two optical fibers multimode/singlemode with four wavelengths at 1310 nm band 8 B 10 B code 300 m – 10 km Frame structure preserved CSMA-CD protocol officially abandoned LAN PHY for local network applications WAN PHY for wide area interconnection using SONET OC-192 c Extensive deployment in metro networks anticipated 635. 412 Spring 2005 Class #3: Ethernet LANs 80

IEEE 802. 3 Family of LAN Protocols The Future of Ethernet? n There appears to be no end to the push for higher speed variants of Ethernet – momentum is building to start work on a 100 Gigabit Ethernet standard! 635. 412 Spring 2005 Class #3: Ethernet LANs 81

IEEE 802. 3 Family of LAN Protocols Appendix: Digital Signal Encoding for 100 BASE-T n n Because of the difficulties associated with higher speed digital communications over UTP, more sophisticated signal encoding schemes are necessary to provide the necessary performance 4 B/5 B-NRZI Encoding – – – This encoding scheme (actually two combined) is used for both 100 BASE-FX and the FDDI specification While Manchester encoding can be used to help maintain synchronization, that reduces the signaling efficiency by 50% and requires the devices to handle a 200 -Mbaud signaling rate; increasing complexity and expense A better way is to use an encoding scheme which has enough redundancy to maintain synchronization but is much more efficient 635. 412 Spring 2005 Class #3: Ethernet LANs 82

IEEE 802. 3 Family of LAN Protocols Appendix: Digital Signal Encoding for 100 BASE-T n Solution: use a 4 B/5 B encoding scheme – 4 bits of ‘data’ are encoded using 5 signal elements n n This increases the efficiency to 80% and requires devices to handle a much lower 125 -Mbaud signaling rate This allows additional symbols to use for low-level link maintenance functions [Table 7. 5 shows all the symbols] – – – Idle signal elements to maintain timing Start & end of frame delimiters (each a pair of signal elements) Transmit error signal element The 4 B/5 B signal elements are encoded again using NRZ-I; differential signaling improves reception in the face of noise To ensure synchronization is maintained the 4 B/5 B signal elements that carry data are chosen (out of the 32 possible elements) to ensure that no element has more than three zeros in a row 635. 412 Spring 2005 Class #3: Ethernet LANs 83

IEEE 802. 3 Family of LAN Protocols Appendix: Digital Signal Encoding for 100 BASE-T n MLT-3 Encoding – – – While 4 B/5 B is great for fiber optic links, it’s spectral characteristics are unsuitable for UTP (radiates a large amount of RF interference) Used with 100 BASE-TX and the twisted pair version of FDDI A ternary voltage encoding scheme that produces a transition for every binary one The 4 B/5 B and NRZ-I encoding schemes are still used, but after that the elements are reconverted to NRZ, scrambled, and then encoded using MLT-3 for transmission Example MLT-3 encoding [Figure 7. 16]: 635. 412 Spring 2005 Class #3: Ethernet LANs 84

IEEE 802. 3 Family of LAN Protocols Appendix: Digital Signal Encoding for 100 BASE-T n MLT-3 Encoding – The State diagram defines how the MLT-3 encoder produces the transmitted ternary symbols [Figure 7. 15] 635. 412 Spring 2005 Class #3: Ethernet LANs 85

IEEE 802. 3 Family of LAN Protocols Appendix: Digital Signal Encoding for 100 BASE-T n 8 B 6 T Encoding – Another ternary encoding scheme used with 100 BASE -T 4 where a block of 8 data bits is encoded into a block of 6 ternary symbols n n n – +V, 0, and –V are three voltage levels used The resulting stream is transmitted round-robin across the three transmit pairs leading to a signaling rate of 25 -Mbaud on each pair The mapping table [Table 7. 6] outlines the 8 B/6 T code translation; the mappings were chosen to maintain proper timing and dc balance (under certain circumstances 6 T code groups can be inverted for transmission to help maintain DC balance) Theoretically similar the 4 B/5 B and other block encoding schemes except ternary symbols are used 635. 412 Spring 2005 Class #3: Ethernet LANs 86

IEEE 802. 3 Family of LAN Protocols Appendix: Digital Signal Encoding for Gigabit Ethernet n 8 B/10 B Encoding – – An encoding scheme similar to 4 B/5 B that was developed by IBM for an earlier mainframe transmission system The developers designed the encoding scheme to provide the following advantages: n n – – Implemented with simple and reliable components Well balanced, which eliminates the need for scrambling Provide sufficient signaling transitions to ensure good synch Provides good error detection capability Though the actual coding is broken into a combination of a 5 B/6 B and a 3 B/4 B encoding scheme, this wasn’t absolutely necessary A process called disparity control can invert code words before transmission to keep the bit stream ‘balanced’ (an equal number of ones and zeros in the symbol stream) 635. 412 Spring 2005 Class #3: Ethernet LANs 87

IEEE 802. 3 Family of LAN Protocols Appendix: Digital Signal Encoding for Gigabit Ethernet n 4 D-PAM 5 Encoding – – A complex scheme used for 1000 BASE-T over twisted pair; required to provide a required data rate with an acceptable level of bit errors The 4 D-PAM 5 has the following key features: n Uses four pair of Category 5 cable; each cable operates in full-duplex mode and carries 250 -Mbps in each direction n PAM 5 (Pulse Amplitude Modulation) is used to send two bits per symbol; keeps the signaling rate below 125 Mbaud n Uses Forward Error Correction to combat crosstalk and noise (that’s why there are five signal levels used) n Scrambling required to randomize signaling stream before transmission 635. 412 Spring 2005 Class #3: Ethernet LANs 88

IEEE 802. 3 Family of LAN Protocols Appendix: Scrambling n Data Transmission and Scrambling – – – A technique for randomizing data because long strings of ones or zeros can cause problems in transmission systems (e. g. – loss of timing) A scrambler consists of a set of feedback registers & exclusive OR operations Different scrambling functions can be used based on the amount of scrambling desired & the complexity of the system n – 100 BASE-TX uses Bm = Am X 9 X 11 At the receiver the same scrambling algorithm is used on the received (scrambled) data stream to recover the ‘real’ data stream 635. 412 Spring 2005 Class #3: Ethernet LANs 89

IEEE 802. 3 Family of LAN Protocols Appendix: Scrambling n Binary Logic Diagram of a Typical Scrambler Circuit 635. 412 Spring 2005 Class #3: Ethernet LANs 90

IEEE 802. 3 Family of LAN Protocols Homework & Reading n Reading – – This week’s material: Stallings chapter 7 Next week: advanced features of Ethernet n n n Full-duplex operation Flow Control Link Aggregation Virtual LANs (VLANs) Priority Transport Port Authentication 635. 412 Spring 2005 Class #3: Ethernet LANs 91