Wide Area Networks

WAN vs LANSpanBWDelayDifferent protocolsUsually you dont own the WAN infrastructure

Point to point linkThats what you seeEx: leased lineUsually simulated by a circuit or packet switched network

Circuit SwitchingBased on the PSTN (Public Switched Telephone Network)Analog: modems up to 56K Digital: 64K circuits - SDH w/ TDMcf BocqDesignated circuits

Packet SwitchingData streams segmented in packetsStatistical Multiplexing (FIFO or QoS techniques)

Circuit vs Packet switchingCircuit: Sum of peak data rates < transmission capacityPacket: Sum of average data rates < transmission capacityCircuit: waste of BWPacket: delay => unacceptable for voice

Connection oriented vs ConnectionlessCircuit: COData: CL => need addressing

Virtual CircuitsConnection Oriented: encapsulation includes a flow identifierBest of two worlds? Switched VCs - 3 phases: circuit setup, data transfer, circuit terminationPermanent VCs - more expensive as need to be constantly up, use less BW

VC multiplexing

Synchronous Data Link ControlSDLC

SDLCDevelopped by IBM for use w/ SNAMost of L2 protocols are based on the SDLC format (HDLC, LAPB, 802.2, etc)

SDLC Frame Format

X.25

X.251970sData Terminal Equipment (DTE)Data Circuit-terminating Equipment (DCE) Packet Switching Exchange (PSE)DCE provides clock

X.25 topology

Packet Assembler/Disassembler

X.25 Stack

LAPB Frame

X.25 Data Link ControlPoint to point full duplex data linksCorrection of errors and congestion controlEncapsulation of data in variable length frames delimited by flagsRedundant error correction bitsSliding window (8 or 128 frames)

X.121 address

X.121 addressData Network Identification Code (DNIC) National Terminal Number (NTN)

Packet Level ProtocolSeveral circuits multiplexedSliding window error and congestion control for every VCCall restriction, charging, QoS, ...

VC SetupPVC: permanent entry in routing table (static), substitute to leased linesSVC: dynamic entry in routing table triggered by an open packet and torn down by close packet

Frame Relay

CharacteristicsIntroduced in 1984 but only (significantly) deployed in the late 1980sL1 and 2Packet Switched technology: PVCs and SVCsConnection-oriented data link layer communicationX.25 lite

Differences with X.25Less robustAssumes more reliable medium => No retransmission of lost dataNo windowingError control handled by higher layersHigher performance and transmission efficiency

Frame Relay Topology

DLCIData Link Connection IdentifierUniquely identify circuitsAssigned by service providerLocal significance only (except with LMI)

DLCI

Frame Format

Discard EligibilityOne bit in the address fieldIdentifies lower importance traffic that will be dropped first if congestion occursSet by DTE equipment

Congestion Control: FECNFECN: Forward Explicit Congestion NotificationDCE sets FECN bit to 1When received by DTE, it indicates that frame experienced congestionSent to higher layers or ignored

Congestion Control: BECNBECN: Backward Explicit Congestion NotificationSame as FECN but set on the return flow

LMILocal Management InterfaceFrame Relay extension Introduced in 1990 by the gang of four (Cisco, DEC, Nortel and Stratacom)Additional capabilities for complex internetworking environmentsLater Standardized by CCITT

LMI (2)Global addressing: DLCIs become global addressesVirtual-circuit status messagesMulticasting

LMI Frame Format

CIRWhat you buy with a FR connectionCommitted Information RateCIR= Committed Burst/Committed TimeAlso Maximum Rate

ATMAsynchronous Transfer Mode

CharacteristicsOriginally designed to transmit voice, video and data over the same networkCell switchingEach communication is assigned a timeslotTimeslots are assigned on a demand-basis => asynchronous (as opposed to TDM)

Cells53 bytes: 5 byte header + 48 byte payloadTradeoff between voice world and data world:Voice needs small payloads and low delayData needs big payload and less overhead

ATM InterfacesUNI: User to Network InterfaceNNI: Network to Network Interface

ATM Interfaces

UNI and NNI cell formats

UNI and NNI differencesNNI has bigger VPI rangeUNI has Generic Flow Control fieldGFC used to identify different end stations

VPI and VCIUsed to determine pathsVPI: Virtual Path IdentifierVCI: Virtual Channel Identifier VPI identifies a bundle of VCIs

VPI and VCI (2)

ATM SwitchingTable look upIncoming interface/VPI/VCI is mapped to an outgoing interface/VPI/VCI

ATM Reference Model

ATM Adaptation Layer (AAL)Together with ATM layer, equivalent to Data Link layer in OSI modelAAL1: Connection Oriented => Voice and VideoAAL 3,4: Connection Oriented and Connectionless (similar to SMDS)AAL 5: Connection Oriented and Connectionless for CLIP and LANE

ATM Sources

ATM AddressesITU-T Standard: E.164 (Telephone #)ATM Forum defined 20-byte NSAP Addresses for use in private networksE.164 address used as prefix on NSAPMapped to IP addresses by ATM ARP (in CLIP)

ATM QoSTraffic Contract: peak bandwidth, average sustained bandwidth, burst size , Similar to FRTraffic Shaping (end device): Queuing, BufferingTraffic Policing (switches): Enforces contract

Path Establishment

LAN Emulation (LANE)Purpose: emulate a LAN over an ATM networkEthernet or Token RingResolves MAC addresses to ATM addresses

LANE Equivalent

LANE ComponentsLEC: LAN Emulation ClientLES: LAN Emulation ServerBUS: Broadcast and Unknown ServerLECS: LAN Emulation Configuration Server

LANE Components

InitializationLEC finds LECS via pre-established ILMI procedure or through well-known circuitLECS returns: ATM address of the LES, type of LAN being emulated, maximum packet size on the ELAN, and ELAN name LEC registers to its LES (LES checks with LECS)LES assigns LECID (LE Client ID)

CommunicationLE ARP Request sent to LESIf LES doesnt know, it floods the request