Reti di Calcolatori - Slide 23

download Reti di Calcolatori - Slide 23

of 99

Transcript of Reti di Calcolatori - Slide 23

  • 7/30/2019 Reti di Calcolatori - Slide 23

    1/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 1/

    Reti di Calcolatori

    Architetture e Protocolli

    Giorgio Ventre

    Dipartimento di Informatica e SistemisticaUniversit di Napoli Federico II

  • 7/30/2019 Reti di Calcolatori - Slide 23

    2/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 2/

    The OSI Reference Model

    The Data Link Layer

  • 7/30/2019 Reti di Calcolatori - Slide 23

    3/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 3/

    The OSI Reference Model: the Data Link Layer

    In a LAN, the Data Link Layer is responsible foroffering access to the network to/from the upper

    layers

    Different technical solutions depending on the type

    of service required by the applications

    Office automation applications might ask forsimplicity and capillarity, while factory automationneeds performance guarantees in data delivery

  • 7/30/2019 Reti di Calcolatori - Slide 23

    4/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 4/

    The OSI Reference Model: LAN Architectures

    The MAP protocol suite Manifacturing Automation

    Protocol Proposed by GM to haveperformance guaranteesin data delivery for

    assembly lines Based on a token-passing

    mechanism to have

    fairness in media access

    The TOP protocol suite Technical and Office

    Protocol Proposed by Boeing as astandard for officeautomation

    Designed for offering wideconnectivity andcommunication to

    computers Based on a contentionmechanism

  • 7/30/2019 Reti di Calcolatori - Slide 23

    5/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 5/

    The OSI Reference Model: LAN Architectures

    The IEEE 802 Standard Needed to provide a standard for LAN manifacturers (1980)

    Compromise over different technical solutions Offers a common interface to the implementation of upper

    layers protocols

    802.3 802.4 802.5 802.6

    802.1 Bridging

    802.2Logical LinkSub-Layer

    MACSub-Layer

    Physical LinkLayer

    DataLinkLayer

    802.9

    802.10 Security & Privacy

    802Overview

    &Architecture

    802.1Management

  • 7/30/2019 Reti di Calcolatori - Slide 23

    6/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 6/

    The OSI Reference Model: LAN Architectures

    IEEE 802.3 - Ethernet (1983) Bus Topology with contention based access

    IEEE 802.4 - Token Bus (1983) Bus Topology with token-passing mechanism

    IEEE 802.5 - Token Ring (1984) Ring Topology with token passing mechanism IEEE 802.6 - DQDB (1990)

    Standard for Metropolitan Area Network

    IEEE 802.9 - New for Isochronous LANs

  • 7/30/2019 Reti di Calcolatori - Slide 23

    7/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 7/

    The OSI Reference Model: LAN Architectures

    The IEEE 802.X Standards have been adoptedor recognized by

    National Bureau of Standards International Standard Organisation (ISO) as the

    8802.X Standard

    European Computer Manifacturer Association

    They have inspired other important standards

    EIA/TIA 568 for Structured Cabling Systems

  • 7/30/2019 Reti di Calcolatori - Slide 23

    8/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 8/

    ISO/OSI, IEEE 802, EIA/TIA 568

    ApplicazionePresentazione

    Sessione

    Trasporto

    Rete

    Data Link

    Fisico

    6

    54

    3

    21

    7

    IEEE 802

    WAN

    EIA/TIA 568 (*)

    (*) in futuro ISO/IEC 11801

    PA

    BX

    .

    .

    .

  • 7/30/2019 Reti di Calcolatori - Slide 23

    9/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 9/

    IEEE 802.1

    IEEE 802.1 lo standard che contiene lespecifiche generali del progetto 802

    IEEE 802.1 uno standard composto da molteparti tra cui:

    802.1 Part A: Overview and Architecture802.1 Part B: Addressing Internetworking andNetwork Management

    802.1 Part D: MAC Bridges

  • 7/30/2019 Reti di Calcolatori - Slide 23

    10/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 10/

    Il Data Link nelle LAN

    Le LAN sono reti di tipo broadcast in cui ognistazione a livello data link riceve i frame inviati

    da tutte le altre stazioni Il data link broadcast pu essere realizzato sia

    con topologie broadcast quali il bus, sia contopologie punto a punto quali l'anello

    I canali trasmissivi sono sufficientemente

    affidabili e non necessario in generecorreggere gli errori a questo livello

  • 7/30/2019 Reti di Calcolatori - Slide 23

    11/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 11/

    I Sottolivelli

    Per tener conto delle precedenti peculiarit ilprogetto IEEE 802 ha suddiviso il livello data link

    in due sottolivelli:LLC: Logical Link ControlMAC: Media Access Control

  • 7/30/2019 Reti di Calcolatori - Slide 23

    12/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 12/

    LLC (Logical Link Control)

    linterfaccia comune a tutte le LAN verso illivello network.

    I servizi e i protocolli di questo sottolivello sonodescritti nello standard IEEE 802.2

  • 7/30/2019 Reti di Calcolatori - Slide 23

    13/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 13/

    MAC (Media Access Control)

    specifico per ogni LAN e risolve il problemadella condivisione del mezzo trasmissivo

    Esistono vari tipi di MAC: ad allocazione dicanale fissa o dinamica, deterministici ostatistici, ecc.

  • 7/30/2019 Reti di Calcolatori - Slide 23

    14/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 14/

    IEEE 802.3 (CSMA/CD)

    Topologia: busCablaggio: bus, stellaArbitraggio del canale trasmissivo: tramite

    contesa

    Tipologia del protocollo: non deterministicoVelocit Trasmissiva: 10 Mb/s

    Throughput massimo: 4 Mb/sEvoluzione della rete Ethernet proposta da

    Digital, Intel, Xerox (DIX).

  • 7/30/2019 Reti di Calcolatori - Slide 23

    15/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 15/

    IEEE 802.4 (Token Bus)

    Topologia: busCablaggio: busArbitraggio del canale trasmissivo: token Tipologia del protocollo: deterministico

    Velocit Trasmissiva: 10 Mb/s Throughput massimo: 8 Mb/s

    Standard di rete utilizzato in ambito di fabbricaspecialmente in relazione al MAP(Manufacturing Automation Protocol)

  • 7/30/2019 Reti di Calcolatori - Slide 23

    16/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 16/

    IEEE 802.5 (Token Ring)

    Topologia: anelloCablaggio: stella o doppio anelloArbitraggio del canale trasmissivo: token Tipologia del protocollo: deterministico

    Velocit Trasmissiva: 4 o 16 Mb/s Throughput massimo: 3 o 12 Mb/s

    Evoluzione della rete Token Ring proposta daIBM in alternativa a Ethernet

  • 7/30/2019 Reti di Calcolatori - Slide 23

    17/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 17/

    ISO 9314

    Topologia: anelloCablaggio: doppio anello o stellaArbitraggio del canale trasmissivo: token Tipologia del protocollo: deterministico

    Velocit Trasmissiva: 100 Mb/s Throughput massimo: 80 Mb/s

    Primo standard per reti locali concepito peroperare su fibra ottica

    (FDDI: Fiber Distributed Data Interface)

  • 7/30/2019 Reti di Calcolatori - Slide 23

    18/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 18/

    Il pacchetto MAC

    I campi principali di un pacchetto a livello MAC

    sono:

    IndirizzoDestinatario

    IndirizzoMittente

    DATI FCS

    DSAP SSAP LLC-PDU CRC

    OTTETTI 6 6 variabile 4

  • 7/30/2019 Reti di Calcolatori - Slide 23

    19/99

  • 7/30/2019 Reti di Calcolatori - Slide 23

    20/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 20/

    Indirizzi MAC

    Si compongono di due parti di 3 Byte: I tre byte pi significativi indicano il lotto di indirizziacquistato dal costruttore della scheda, detto

    anche vendor code o OUI (Organization UniqueIdentifier). I tre meno significativi sono una numerazione

    progressiva decisa dal costruttore

    Esempio: una scheda con MAC address 08-00-2b-3c-07-9a una scheda prodotta da Digital inquanto il lotto 08-00-2b stato acquistato daDigital

    Sono di tre tipi:Single: indirizzo di una singola stazioneMulticast: indirizzo di un gruppo di stazioniBroadcast: indirizzo di tutte le stazioni

  • 7/30/2019 Reti di Calcolatori - Slide 23

    21/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 21/

    Indirizzo di Broadcast

    Il broadcast un tipo particolare di multicast cheindica tutte le stazioni sulla rete locale

    La sua codifica esadecimale ff-ff-ff-ff-ff-ff

  • 7/30/2019 Reti di Calcolatori - Slide 23

    22/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 22/

    Ricezione

    Scheda di Rete Locale

    Rete locale

    Transceiver

    Quando la scheda LAN

    deve passare la tramaricevuta alla CPU?

    Transceivercable

  • 7/30/2019 Reti di Calcolatori - Slide 23

    23/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 23/

    Ricezione

    Ogni scheda di rete, quando riceve unpacchetto, lo passa ai livelli superiori nei

    seguenti casi:Broadcast: sempreSingle: se il DSAP uguale a quello hardware della

    scheda (scritto in una ROM) o a quello caricato dasoftware in un apposito buffer

    Multicast: se stato abilitato via software

  • 7/30/2019 Reti di Calcolatori - Slide 23

    24/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 24/

    Indirizzi di Gruppo

    Servono tipicamente per scoprire i nodiadiacenti

    Esistono due modi diversi di impiego:Solicitation: la stazione che interessata a scoprire

    chi offre un dato servizio invia un pacchetto dimulticast allindirizzo di quel servizio. Le stazioni cheoffrono il servizio rispondono alla solicitation

    Advertisement: le stazioni che offrono un serviziotrasmettono periodicamente un pacchetto dimulticast per informare di tale offerta tutte le altrestazioni

  • 7/30/2019 Reti di Calcolatori - Slide 23

    25/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 25/

    Primitive di Servizio

    A livello MAC esistono solo tre delle quattroprimitive presenti ai livelli superiori:

    Request IndicationConfirm

    La Response non usataConfirm viene generata o dal protocollo locale,

    come indicazione dell'avvenuta trasmissione, odal protocollo remoto, come indicazionedell'avvenuta ricezione.

  • 7/30/2019 Reti di Calcolatori - Slide 23

    26/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 26/

    Primitive di Servizio

    LLC LAYER MACLAYER

    CORRESPONDENTLLC LAYER

    MA.DATA.requestMA.DATA.confirm

    MA.DATA.request

    MA.DATA.indication

    MA.DATA.indication

    TIMEMA.DATA.confirm

    (a)

    (b)

    (a) usato da 802.3

    (b) usato da 802.4, 802.5, FDDI

  • 7/30/2019 Reti di Calcolatori - Slide 23

    27/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 27/

    LAN Architectures: Medium Access Control

    The Medium Access Control Problem Where can we apply the control ?

    Distributed Control Centralised Control

    How can we enforce the control ?

    Sinchronous techniques Asinchronous techniques

    Which type of control can we apply ?

    Continuous Traffic Bursty Traffic

    Are the assumptions still valid today ?

  • 7/30/2019 Reti di Calcolatori - Slide 23

    28/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 28/

    LAN Architectures: Medium Access Control

    Reservation Static or dynamic channel assignment Centralised or distributed control

    Suited for continuous traffic Round Robin

    Static or dynamic cyclic channel assignment

    Centralised or distributed control For bursty and continuous traffic

    Contention

    Inherently distributed (local) control Dynamic channel assignment Suited for bursty traffic

  • 7/30/2019 Reti di Calcolatori - Slide 23

    29/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 29/

    LAN Architectures: Medium Access Control

    Carrier Sensing Multiple Access (CSMA) Very simple model: multiple stations connected

    through a shared media (bus) Random Access Contention

    Derived from the ALOHA Protocols Two-state behavior of a system

    Thinking

    Transmitting Contention is perceived from destruction/garbling of

    data on shared media

  • 7/30/2019 Reti di Calcolatori - Slide 23

    30/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 30/

    LAN Architectures: Medium Access Control

    Non persistent If no carrier, transmit

    If channel occupied, wait time T=f(Prob) 1-Persistent

    If no carrier, transmit

    If channel occupied, wait until again free If collision (no ack), wait time T=Random

    P-Persistent If no carrier, transmit with probability P If channel occupied, wait until again free

  • 7/30/2019 Reti di Calcolatori - Slide 23

    31/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 31/

    LAN Architectures: Medium Access Control

    CSMA suffers of inefficiency due to the problem ofdetecting collisions

    The effective channel bandwidth is then reduced by theexistence of a contention state for the network

    The duration of this state is linked to the

    Frame dimension Technique used to detect collisions

    Frame Frame Frame Frame

    Contention Slots Idle Contention Interval

    LAN A hi M di A C l

  • 7/30/2019 Reti di Calcolatori - Slide 23

    32/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 32/

    LAN Architectures: Medium Access Control

    CSMA/Collision Detect Collisions are detected by listening over the media

    while transmitting a frame If no carrier, transmit If channel occupied, wait until again free

    If collision, stop and transmit collision warning Wait for time T=f(Random Backoff)

    The problem is now ensuring that

    collisions can be detected all the stations can hear the collision warning the duration of the collision state be shortest possible

    LAN A hi E h

  • 7/30/2019 Reti di Calcolatori - Slide 23

    33/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 33/

    LAN Architectures: Ethernet

    Ethernet Developed by Xerox PARC (Metcalfe-Boggs) in late

    70 with contribution from Intel and DEC High reliability (no critical dependecies on anyone of

    the network components)

    Low installation and operation costs Distributed control Suited for data (burst) traffic Fits the typical office environment

    CSMA/CD

    LAN A hit t Eth t

  • 7/30/2019 Reti di Calcolatori - Slide 23

    34/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 34/

    LAN Architectures: Ethernet

    Coding should allowdetection of collision(Manchester)

    Frames should be longenough to ensure collisiondetection

    A maximum cable length isset to avoid non detectionsdue to attenuation

    The longer the frame, the

    longer the collision state Binary Exponential Back-Off

    to reduce collision congestion

    A B

    A B

    A B

    A Bt = 2 T

    LAN A hit t Eth t

  • 7/30/2019 Reti di Calcolatori - Slide 23

    35/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 35/

    LAN Architectures: Ethernet

    Preamble

    Start of Frame (SFD)

    Destination Addr.

    Source Addr.

    Length

    Data

    Pad

    Frame Check

    7 Bytes

    1 Byte

    6 Bytes

    6 Bytes

    2 Bytes

    < 1518 Bytes

    4 Bytes

    Operational Parameters Bit Rate 10 Mbps Slot Time 512 Bit/t Interframe gap 9.6 sec

    Backoff limit 10 Attempt limit 16 Jam size 32 bits Maximum Frame 1518 bytes

    Minimum Frame 512 bytes

    The bus topology allowsmulticasting

    Destination Addr. is a Group Addr.to be programmed in the stationsbroadcasting

    Destination Addr. is 111111

    LAN Architectures Ethernet

  • 7/30/2019 Reti di Calcolatori - Slide 23

    36/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 36/

    LAN Architectures: Ethernet

    SIGNALRX ?

    CARRIER = ONSYNC & WAIT SFD

    START RX

    FCS & SIZE?

    SEND FRAMEUP

    DROP FRAME

    N

    Y

    N

    N

    Y

    DEST_ADD?

    Y

    WAIT PACKETFORM FRAME

    CARRIERON ?

    WAIT FRAME GAPTX FRAME

    COLLISION?

    COMPLETE TXTX = OK

    ATTE_MAX = TRUETX = NOK

    TX JAMATTE ++

    MAX ATTE?

    BACKOFF = R(ATT)WAIT BACKOFF

    Y

    N

    Y

    N

    N

    Y

    TXAlgorithm

    RXAlgorithm

    I mezzi trasmissivi IEEE 802 3

  • 7/30/2019 Reti di Calcolatori - Slide 23

    37/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 37/

    I mezzi trasmissivi IEEE 802.3

    Tipo diSegmento

    Num Max.di MAU

    LunghezzaMassima

    Min. velocitdi propagaz.

    10Base5

    10Base2FOIRL10baseFL10baseFB

    10BaseT

    100

    30222

    2

    500

    185100020002000

    100

    0.77 c (*)

    0.65 c0.66 c0.66 c0.66 c

    0.59 c

    2165

    95050001000010000

    565

    RitardoMax. (ns)

    * : c = 3 x 108 m/s (velocit della luce nel vuoto)

    LAN Architectures: Ethernet

  • 7/30/2019 Reti di Calcolatori - Slide 23

    38/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 38/

    LAN Architectures: Ethernet

    How Ethernet performs ? The random-based contention makes impossible to get any

    bounds on the transmission delay induced However, a worst case analysis can help us in figuring out

    how efficient can be an Ethernet High, constant workload from Kstations

    Constant transmission probability p(no Exp. backoff)

    If A is the probability that the net is alreadyacquired in a slot, then

    A = K p (1 - p)k-1

    For p = 1 / Kthen A = A max = (1 - 1/K) k-1

    So A tends to 1 if Ktends to infinite

    LAN Architectures: Ethernet

  • 7/30/2019 Reti di Calcolatori - Slide 23

    39/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 39/

    LAN Architectures: Ethernet

    The probability a contention is made ofjslots isA (1-A)

    j-1

    and the mean number of slots per contention isSUMj jA (1-A)

    j-1 = 1/A

    A slot has duration 2 then mean contention interval wisw =2 /

    If Pis the mean frame duration, then we can define theEfficiency Eas

    E = P / (P + w) = P / (P +2 / A)

    The longer the cable, the longer the contention interval, thesmaller the Efficiency!

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    40/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 40/

    LAN Architectures: Token Bus

    IEEE 802.4 - Token Bus For automation of assembly lines it is essential to

    have deterministic bounds on the worst casetransmission delay Dmax A token passing mechanism over a ring seems a

    simple solution, since with N stations and for Tframe transmission time, we haveDmax = NT

    However, a ring topology is very sensitive tobreakage and does not fit well an assembly linearchitecture.

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    41/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 41/

    LAN Architectures: Token Bus

    The solution is to define a ring-based logical topologyover a bus architecture

    Each node has a Predecessor and a Successor

    A node passes over the token by broadcasting it on thebus where it is taken by its Successor

    P = BS = C

    P = DS = A

    P = AS = E

    P = ES = B

    P = CS = D

    A B C

    D E

    P = PredecessorS = Successor

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    42/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 42/

    LAN Architectures: Token Bus

    75 Broadband coax

    Bus or Tree structure Transmission:

    Modified Basebad

    (Carrierband) Broadband

    Requires modulation

    equipment (sometime inDTE)

    MAC

    ProtocolFirmware

    Mod/Dem& Control

    Physical

    InterfaceModule

    DTE

    MAC

    Architectural choicesinfluenced by operativeenvironment (factory)

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    43/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 43/

    LAN Architectures: Token Bus

    As in baseband, carrierbandoccupies all the cablebandwidth (1 - 5 - 10 Mbps)

    Signal is modulated phase-coherent frequency shift withno phase change from 1 to 0

    2 Frequencies for 1 and 0 + 3frequencies for control

    All other frequencies can befiltered out (noise)

    Binary 1

    Binary 0

    1 Bit time

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    44/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 44/

    LAN Architectures: Token Bus

    A token based MAC is very sensitive to tokencorruption and station faults

    The MAC protocol must manage events such as Stations joining or leaving the ring Faulty or powering down stations Corrupted tokens/frames

    There is a Tmax maximum

    waiting time for DTEsTmax = 2 * (TxD + ProcD) +

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    45/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 45/

    Preamble

    Start Delimiter (SD)

    Destination Addr.

    Data

    CRCEnd Delimiter

    => 1 Byte

    1 Byte

    1 Byte

    6 Bytes

    6 Bytes

    < 8191 Bytes

    1 Bytes

    Frame Control specifies if a framecontains data or control infoIf data, contains priority and ackgrant information

    If control, contains message type:Token passingToken claim

    (for ring initialization)Solicit successor

    (station joining - periodic)Who follows me

    (lost token)Resolve contention

    (multiple new stations)Set successor

    (station informs predec.about leaving)

    Source Addr.

    Frame Control

    4 Bytes

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    46/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 46/

    Basic network operation1) Wait for token2) Transmit Nmax frames in queue3) Pass token to successor4) Lissen new traffic

    4.1) If valid frames, then my tx was ok4.2) If frames not valid, then there is a problem

    4.2.1) Rigenerate new token

    4.2.2) If still nothing, my successor is dead4.2.3) Who follows me ?4.2.4) Receive Set successor & Synchronize Net

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    47/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 47/

    Priority operation Four Access Classes

    Class 6: Urgent Messages, Alarms Class 4: Control and Management Class 2: Routine Operations

    Class 0: Low priority (e.g. file transfer, downloading) Transmission regulated by two timers

    Token Hold Timer (THT)

    Maximum Transmission Time for a Station High-Priority THT (HP-THT)Maximum Transmission Time for Class 6 Packets

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    48/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 48/

    When the token is received, the station transmits Class6 packets until HP-THT

    Then, if THT has not expired, it is set to a timer calledToken Rotation Timer (TRT) measuring the time

    expired since the last token reception

    The station starts transmitting lower priority packets

    until the difference between THT and a fixed valuecalled Target Token Rotation Time (TTRT) is positive

    LAN Architectures: Token Bus

  • 7/30/2019 Reti di Calcolatori - Slide 23

    49/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 49/

    B

    B

    B

    HP-THT > 0THT > 0

    HP-THT = 0THT := TRT

    HP-THT = 0

    THT = TTRT

    TRT = t

    TRT = 1

    TRT = 2

    Token

    Class 6

    Classes 0-4

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    50/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 50/

    IEEE 802.5 Token Ring Standard originated from/compatible with IBM LAN

    Based on a ring topology Stations connected through a serial line, not via a

    bus

    Media Access Control based on a token mechanism Bandwidth achievable 4 - 16 Mbps Star topology possible with concentrators

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    51/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 51/

    A

    D

    C

    B

    F

    A

    D

    C

    B

    T

    A receivesthe Token

    A

    D

    C

    B

    F

    A tx aFrame to D

    Frame returnsto Athrough D andis eliminated

    A

    D

    C

    B

    T

    A releasesthe Token to B Early Token

    Release

    A

    D

    C

    B

    F

    A tx aFrame to D andreleases immediatelythe token

    F

    T

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    52/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 52/

    The network media is a 2 pairs shielded cable The signal coding is Differential Manchester

    The physical architecture can be a point-to-point ring orthrough a concentrator

    MAC

    ProtocolFirmware

    TrunkCouplingUnit

    DTE

    MACRTR R RTT T

    HUB

    Drop Connection

    Ring In Ring Out

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    53/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 53/

    Since the ring is realised through point-to-point lines,the TCU must guarantee data by-passing when astation is inactive.

    When it is not transmitting, a station is inlistening mode - 1 bit copy mechanism

    Listening Mode

    1 Bit Buffer / Copy Delay

    In Out

    DTE

    Tx Mode

    In Out

    DTE

    To/From Upper Layers

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    54/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 54/

    Start Delimiter

    Access Control

    Destination Addr.

    Data

    CRC

    End Delimiter

    1 Byte

    1 Byte

    1 Byte

    6 Bytes

    6 Bytes

    < 5000 Bytes

    1 Bytes

    Source Addr.

    Frame Control

    4 Bytes

    Data Frame Format

    Start Delimiter

    Access Control (AC)

    => 1 Byte

    1 Byte

    1 ByteEnd Delimiter

    Token Frame Format

    Frame Status 1 Bytes

    Access Control:Token, Monitor, Priority, Reserv.

    Frame Control:For ring management

    Frame Status:A bit - On, Destination OK

    C bit - On, Frame OK

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    55/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 55/

    Token Ring has also a priority mechanism: Each token has a priority assigned

    When a station receives a token with priority P, itcan only transmit frames with priority F > = P

    A station can book a token with a certain priority Mechanisms exists in order to avoid abuses This priority mechanism is not fair as the one

    adopted in Token Bus

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    56/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 56/

    Each MAC Layer maintains two sets of values: Priorities:

    Pm: Max priority among the waiting frames Pr: Priority of the previous token Rr: Reservation contained in the previous token

    Stacks: Sr: Stack for the previous Pr Sx: Stack for new priorities P

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    57/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 57/

    MAC assignes priority to any tx frame P = Current token priority Pr

    R = 0 (to be set by other stations) MAC generates a new token with:

    P = Pr, R= Max(Rr,Pm) P constantif there are no more transmittable frames or noreservation request with higher P

    P = Max(Rr,Pm), R = 0 P increasesif there are still transmittable frames pending orthere is a reservation request for higher priority

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    58/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 58/

    We now need a mechanism to lower the priority:

    The station who raises P becames thestacking station, i.e. a station keeping track of P This station will be responsible for

    raising the priority for new reservations keeping it the same lowering it, as soon as new reservations ask for lower

    priorities

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    59/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 59/

    Ring Management - Inizialization When switched on, a station makes a Duplication Address

    Test (DAT)

    If OK, it send out a Standby Monitor Present (SMP) to informthe immediate downstream station about its existence.

    Ring Management - Standby Monitor

    Each ring has a monitor station, to check token passing andtransmitting periodic Active Monitor Present (AMP) signals In case there are no AMP, a station can send a Claim Token

    (CT) frame to become the monitor (FCFS approach)

    LAN Architectures: Token Ring

  • 7/30/2019 Reti di Calcolatori - Slide 23

    60/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 60/

    Ring Management - Active Monitor When a station becomes the new monitor, it starts a clock

    and sends out a Purge (PRG) frame to eliminate tokens and

    frames It then sends out a AMP signal to inform the other stations,and creates a new token

    The other stations move to a Standby Monitor State

    Ring Management - Beaconing When a station receives no AMP or frames in an interval, it

    transmits a Beacon (BCN) signal to verify the existence of aconnection

    If the signal is not received back from the station, then thering is interrupted somewhere

    LAN Architectures: High Speed LANs

  • 7/30/2019 Reti di Calcolatori - Slide 23

    61/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 61/

    High Speed Lans The Performance Problem -

    More stations More applications More usage

    More media More capillarity

    Demand for More Bandwidth

    This must cope with the technological limitations ofsome LAN architectures

    LAN Architectures: High Speed LANs

  • 7/30/2019 Reti di Calcolatori - Slide 23

    62/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 62/

    An example: improving Ethernet performance If Tc is the delay to detect a collision, and Tix the

    delay to transmit a frame, whenTc

  • 7/30/2019 Reti di Calcolatori - Slide 23

    63/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 63/

    Case 1:Tp = 25 sec (Lenght = 2.5 Km)Fs = 10.000 bits (Frame size)

    B = 10 Mbps (Bandwidth) Tp / Tix = 25 / 1000

    Tc = 50 sec N = Tc x B = 500 (Bits wasted if collision)

    Case 2: B = 100 Mbps, N = 5000 Case 3: B = 200 Mbps, N = 10000 (Frame)

    CSMA/CD is not suited!

    LAN Architectures: High Speed LANs

  • 7/30/2019 Reti di Calcolatori - Slide 23

    64/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 64/

    FDDI : ISO 9314 Token based MAC

    Fibre based (Copper based available - CDDI) Ring 1: Primary Ring Ring 2: Secondary Ring - Backup / Add. Tx

    Single Attachment Stations (SAS) Dual Attachment Stations (DAS)

    Longer distances - Higher speed

    LAN Architectures: High Speed LANs

  • 7/30/2019 Reti di Calcolatori - Slide 23

    65/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 65/

    M

    S

    M

    S

    A B

    A

    BA

    B

    Optical CouplingUnit

    SAS

    DAS

    Wiring Concentrator

    Media InterfaceConnector

    Primary Ring

    Secondary Ring

    Primary Ring

    Master Key

    Slave Key

    LAN Architectures: High Speed LANs

  • 7/30/2019 Reti di Calcolatori - Slide 23

    66/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 66/

    FDDI is used as a High-Speed LAN for clustering top-end computing

    systems (NOW)

    as a High-Speed backbone for LAN interconnection

    FDDI

    Local Area Network

    Bridge

    Campus Network

    BuildingBackbone

    LAN Architectures: High Speed LANs

  • 7/30/2019 Reti di Calcolatori - Slide 23

    67/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 67/

    Operation Mode At any time, only one ring is active and provides

    timing 4B/5B Manchester Coding MAC based on ETR - Token Ring

    Priority mechanism similar to Token Bus Isochronous FDDI defined but not standard

    based on allocation of time-slots to isochronous

    media - Synchronous Allocation Time (SAT)

    LAN Architectures: Bridging

  • 7/30/2019 Reti di Calcolatori - Slide 23

    68/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 68/

    The diffusion of LANs brings in severalproblems:

    How to connect multiple LANs with similartechnologies How to divide a single LAN into separate entities

    How to connect multiple LANs with differenttechnologies

    How to limit and control the traffic flowing from LAN

    to LAN (load balancing and security)

    LAN Architectures: Bridging

  • 7/30/2019 Reti di Calcolatori - Slide 23

    69/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 69/

    Backbone LAN

    Buiding C

    LAN 3

    Buiding D

    LAN 4

    Bridge

    Office LAN

    ComputerBuiding A

    LAN 1 Buiding B

    LAN 2

    Bridging keeps local all the frames to systems in the sameLAN and propagates only frames for external systems

    LAN Architectures: Bridging

  • 7/30/2019 Reti di Calcolatori - Slide 23

    70/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 70/

    Connecting LAN

    Buiding A

    LAN 3

    BridgeOffice LAN

    Computer

    Buiding B

    LAN 2

    Bridging is a solution also for the existing limitationsover the extension of a LAN

    2.5 Km

    2.5 Km

    2.5 Km

    LAN Architectures: Bridging

  • 7/30/2019 Reti di Calcolatori - Slide 23

    71/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 71/

    Backbone LAN

    Buiding C

    LAN 3

    Buiding D

    LAN 4

    Bridge

    Office LAN

    ComputerBuiding A

    LAN 1 Buiding B

    LAN 2

    Bridging keeps local all the problems due tofaulty networks or systems

    Missing

    Terminator

    Faulty

    Transceiver

    Bridging Operation

  • 7/30/2019 Reti di Calcolatori - Slide 23

    72/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 72/

    NET

    LLC

    MAC

    PHY

    NET

    LLC

    MAC

    PHY

    PKT

    802.3 PKT

    802.3 PKT

    PKT

    PKT

    802.3 PKT

    802.3 PKT

    PKT

    PKT

    802.4 PKT

    802.4 PKT

    802.4 PKT

    802.4 PKT

    CSMA/CD Lan 802.4 Lan

    Bridge

    Host A Host B

    Differences in formats and protocols make bridginga complex task.

    Bridging Operation

  • 7/30/2019 Reti di Calcolatori - Slide 23

    73/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 73/

    A long list of problems Different frames:

    translation fragmentation

    Different data rates:

    buffering timing control from faster to slower

    Different MACs

    token vs. collision with different efficiency Different maximum frame length

    no solution!

    Bridging Disadvantages

  • 7/30/2019 Reti di Calcolatori - Slide 23

    74/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 74/

    Store&Forward behaviour with additional delayintroduced

    No flow-control at the MAC sublayer, withpossible buffer overflow when the output LAN isoverloaded

    Differences in frame types demand for frameconversion, with possible multiple errors

    introduced during frame relaying

    Bridging Operation

  • 7/30/2019 Reti di Calcolatori - Slide 23

    75/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 75/

    802.3 to 802.3: only load differences 802.4 to 802.3

    Priority bits missing (bridge lie) Acknowledgement request (bridge lie)

    802.5 to 802.3 A&C bits missing (bridge lie) Priority bits (bridge lie)

    802.4 to 802.4 Acknowledgement request (bridge lie)

    802.5 to 802.4

    A&C bits Priority 802.5 to 802.5

    A&C bits

    Bridging Operation

  • 7/30/2019 Reti di Calcolatori - Slide 23

    76/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 76/

    Transparent bridgingthe presence of one or multiple bridges is

    transparent to the stations Bridges automatically initialize and configure

    dynamically with no external intervention A bridge can have two or more bridge ports. In

    this last case it is called a multiportbridge.

    Bridges operate in promiscuous mode, i.e. theyreceive and buffer all frames received

    Bridging Operation

  • 7/30/2019 Reti di Calcolatori - Slide 23

    77/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 77/

    A bridge includes a MAC chipset with a portmanagement software

    MACChipset

    MACChipset

    MemoryBuffers

    PortMngmt

    ProtocolEntity

    Forw.DataBase

    Station

    Address

    Port

    Num.

    Port 1 Port 2

    Bridge learning

  • 7/30/2019 Reti di Calcolatori - Slide 23

    78/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 78/

    The Forwarding Data Base (FDB) is formed bydynamic learning and interaction with otherbridges A newly inserted bridge has an empty FDB For each received frame, the source address and

    the entry port are stored in the FDB

    To ensure delivery, the frame is forwarded over allthe ports (flooding)

    The procedure is repeated for each frame from anewsource

    OK for static scenarios: no moving stations/links

    Bridge learning

  • 7/30/2019 Reti di Calcolatori - Slide 23

    79/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 79/

    The Inactivity Timer (IT) is a value in the FDB thatrecords the silence time of a source.

    When IT > Tmax the entry is removed

    In this way the FDB is formed only of active stationsand has a limited size

    A spanning tree algorithm is used to evaluate theproper forwarding route in case of multiple paths

    In large networks, stations can move and links can

    change state so a new spanning tree must be created

    Topology Initialization

  • 7/30/2019 Reti di Calcolatori - Slide 23

    80/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 80/

    Bridges exchange regularly special frames calledBridges Protocol Data Units (BPDU). These frames arenot directly forwarded, but used to produce new

    BPDUs Each bridge has an Id and a priority. The root bridge

    has the highest priority and smallest Id

    A new bridge starts to operate in the rootstate andtransmists over all ports the Configuration BPDUs

    Id of the current root (itself initially) Path cost to the root (0 initially) Id of the transmitter Id of the port

    Topology Initialization

  • 7/30/2019 Reti di Calcolatori - Slide 23

    81/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 81/

    When a bridge receives a BPDU it compares the rootPriority and Id with its own values.

    If this is the case, the bridge sets the new value for theroot and computes the new cost for the root path

    A new Configuration BPDU is formed and forwarded toall the other ports

    By knowing the costs to the root for each port, a bridgeis designatedto be its way to the root. It will be the onlyfrowarding bridge in that segment

    The Id value is again the tie breaker value

    Topology change

  • 7/30/2019 Reti di Calcolatori - Slide 23

    82/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 82/

    The root bridge will transmit regularly ConfigurationBPDUs. These are processed by the bridges on thepath to inform all other bridges about changes

    A Message Age Timer is kept for each port. In case ofa port/link failure, the MAT expires and a procedure isactivated to change topology

    Topology Change Notification BPDUs are exchangedto inform the root about port status modifications Exists a procedure also to change the root and the

    designated bridge

    Il progetto IEEE 802: il sottolivello LLC

  • 7/30/2019 Reti di Calcolatori - Slide 23

    83/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 83/

    FDDI802.3 802.4 802.5

    FDDI

    802.2 Logical Link ControlISO 8802.2

    802.6

    LIVELLONETWORK

    LIVELLO

    DATA LINK

    LIVELLOFISICO

    LLC

    MAC

    CSMA/CD TOKENBUS

    TOKENRING

    DQDB

    Interfaccia unificata con il livello network

    Tecnologie trasmissive differenziate

    ISO8802.3

    ISO8802.4

    ISO8802.5

    ISO8802.6

    ISO9314

    (Local and Metropolitan Area Network)

    Il sottolivello LLC

  • 7/30/2019 Reti di Calcolatori - Slide 23

    84/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 84/

    Definisce i protocolli usati per realizzare una opi connessioni logiche su di un singolo mezzofisico

    LLC deve essere utilizzabile con uno qualunquedei MAC

    PDU del LLC

  • 7/30/2019 Reti di Calcolatori - Slide 23

    85/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 85/

    DESTINATION

    ADDRESS

    SOURCE

    ADDRESS

    CONTROL INFORMATION

    1 1 1 o 2 variabileOTTETTI

    Il campo control permette lesistenza di tre tipi dipacchetti:

    Unnumbered Supervisor Information

  • 7/30/2019 Reti di Calcolatori - Slide 23

    86/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 86/

    Livello 3

    OSI

    Livello 3

    TCP/IP

    Altro

    livello 3

    Sottolivello MAC

    Sottolivello LLC

    Scelta basatasu LLC-DSAP

    SAP di LLC

  • 7/30/2019 Reti di Calcolatori - Slide 23

    87/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 87/

    Servono ad indicare quale protocollo di livellosuperiore ha originato il pacchettoSono grandi un Byte

    due bit I/G e U riservati64 indirizzi singoli, globali definibili

    ff broadcast00 data link layer itself

    1 0 1 1 1 1 01

    I/G (Individual=0, Group=1)U (Non Universal=0, Universal=1)

    SAP LLC Universal

  • 7/30/2019 Reti di Calcolatori - Slide 23

    88/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 88/

    Gli indirizzi globali sono assegnati dallISO soloper i protocolli progettati da un comitato distandardizzazione

    Esempi di SAP-LLC

  • 7/30/2019 Reti di Calcolatori - Slide 23

    89/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 89/

    La codifica 0FEH indica il protocollo ISO 8473Internet Protocol

    La codifica 042H indica il protocollo IEEE802.1D Spanning Tree Configuration La codifica 0AAH indica un pacchetto LLC

    speciale detto SNAP

    0FEH 0FEH UI

    1 1 1 mOTTETTI

    DSAP SSAP CONTROL INFO

    Servizi LLC

  • 7/30/2019 Reti di Calcolatori - Slide 23

    90/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 90/

    LLC offre al Livello 3 i seguenti tipi di servizio:Tipo 1: Unacknowledged Connectionless Service

    Tipo 2: Connection Oriented ServiceTipo 3: Semireliable Service

    LLC Tipo 1

  • 7/30/2019 Reti di Calcolatori - Slide 23

    91/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 91/

    Unacknowledged Connectionless ServiceServizio non connesso

    Non esiste un acknowledge (ACK)Se il messaggio perso o rovinato dai disturbi

    non viene fatto nessun tentativo, a questo livello,

    di recuperare l'erroreNessuna procedura di controllo di flusso: le

    trame non sono numerate La trasmissione pu essere punto-punto,

    multipunto o broadcast

    Primitive per LLC tipo 1

    Q i i LLC l li i hi d

  • 7/30/2019 Reti di Calcolatori - Slide 23

    92/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 92/

    Questo servizio LLC molto semplice e richiedesolo due primitive:L.DATA.requestL.DATA.indication

    USER LLCLAYER

    CORRESPONDENTUSER

    L.DATA.request

    L.DATA.indication

    LLC tipo 2

    S i i

  • 7/30/2019 Reti di Calcolatori - Slide 23

    93/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 93/

    Servizio connesso il servizio pi sofisticato che il livello data link

    possa offrire al livello networkSorgente e destinazione aprono una connessione

    prima di trasferire i dati e la chiudono al termine

    I frame sono numerati e il livello 2 garantisce che:ogni frame inviato sia ricevuto correttamenteogni frame sia ricevuto esattamente solo una volta

    tutti i frame siano ricevuti nell'ordine correttoEsistono meccanismi di controllo di flusso (flow

    control)

    Primitive LLC tipo 2

    I t i d ll C i

  • 7/30/2019 Reti di Calcolatori - Slide 23

    94/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 94/

    Instaurazione della Connessione:L.CONNECT.requestL.CONNECT.indicationL.CONNECT.confirm

    Trasferimento dei dati:L.DATA_CONNECT.requestL.DATA_CONNECT.indicationL.DATA_CONNECT.confirm

    Chiusura della Connessione:L.DISCONNECT.requestL.DISCONNECT.indicationL.DISCONNECT.confirm

    LLC tipo 2

    R t d ll C i

  • 7/30/2019 Reti di Calcolatori - Slide 23

    95/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 95/

    Reset della Connessione:L.RESET.request

    L.RESET.indicationL.RESET.confirm

    Flow control:

    L.FLOWCONTROL.requestL.FLOWCONTROL.indication

    LLC tipo 2

    Il t l l 1 2 B t

  • 7/30/2019 Reti di Calcolatori - Slide 23

    96/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 96/

    Il campo control pu essere lungo 1 o 2 Byte Il campo control lungo 2 Byte:

    I: information il pacchetto di dati. I numeri disequenza sono lunghi 7 bitRR: Receiver Ready un ACK frame quando non

    esiste traffico per piggybackingRNR: Receiver Not Ready come RR, ma invita il

    trasmettitore a sospendere la trasmissione

    REJ: Reject indica un errore di trasmissione. Ilmittente deve ritrasmettere tutti i pacchetti a partireda quello errato

    The ultimate winner in the LAN war

  • 7/30/2019 Reti di Calcolatori - Slide 23

    97/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 97/

    The ultimate winner in the LAN war

  • 7/30/2019 Reti di Calcolatori - Slide 23

    98/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 98/

    The ultimate winner in the LAN war

  • 7/30/2019 Reti di Calcolatori - Slide 23

    99/99

    Dipartimento di Informatica e Sistemistica11/05/2007 GV/RC/A&P 99/