Wide Area Networks

uptightexampleNetworking and Communications

Oct 24, 2013 (3 years and 7 months ago)

66 views

Wide Area Networks

WAN vs LAN


Span


BW


Delay


Different protocols


Usually you don’t own the WAN
infrastructure

Point to point link


That’s what you “see”


Ex: leased line


Usually simulated by a circuit or packet
switched network

Circuit Switching


Based on the PSTN (Public Switched
Telephone Network)


Analog: modems up to 56K


Digital: 64K circuits
-

SDH w/ TDM


cf Bocq


Designated circuits


Packet Switching


Data streams segmented in packets


Statistical Multiplexing (FIFO or QoS
techniques)



Circuit vs Packet switching


Circuit: Sum of peak data rates <
transmission capacity


Packet: Sum of average data rates <
transmission capacity


Circuit: waste of BW


Packet: delay => unacceptable for voice

Connection oriented vs
Connectionless


Circuit: CO


Data: CL => need addressing

Virtual Circuits


Connection Oriented: encapsulation
includes a “flow” identifier


Best of two worlds?


Switched VCs
-

3 phases: circuit setup, data
transfer, circuit termination


Permanent VCs
-

more expensive as need to
be constantly up, use less BW

VC multiplexing

Synchronous Data Link Control


SDLC

SDLC


Developped by IBM for use w/ SNA


Most of L2 protocols are based on the
SDLC format (HDLC, LAPB, 802.2,
etc…)


SDLC Frame Format

X.25

X.25


1970s


Data 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 Control


Point to point full duplex data links


Correction of errors and congestion control


Encapsulation of data in variable length
frames delimited by flags


Redundant error correction bits


Sliding window (8 or 128 frames)

X.121 address

X.121 address


Data Network Identification Code (DNIC)


National Terminal Number (NTN)


Packet Level Protocol


Several circuits multiplexed


Sliding window error and congestion
control for every VC


Call restriction, charging, QoS, ...

VC Setup


PVC: permanent entry in “routing” table
(static), substitute to leased lines


SVC: dynamic entry in “routing” table
triggered by an “open” packet and torn
down by “close” packet

Frame Relay

Characteristics


Introduced in 1984 but only (significantly)
deployed in the late 1980s


L1 and 2


Packet Switched technology: PVCs and
SVCs


C
onnection
-
oriented data link layer
communication


X.25 “lite”

Differences with X.25


Less robust


Assumes more reliable medium =>



No retransmission of lost data


No windowing


Error control handled by higher layers


Higher performance and transmission
efficiency


Frame Relay Topology

DLCI


Data Link Connection Identifier


Uniquely identify circuits


Assigned by service provider


Local significance only (except with LMI)


DLCI

Frame Format

Discard Eligibility


One bit in the address field


Identifies lower importance traffic that will
be dropped first if congestion occurs


Set by DTE equipment

Congestion Control: FECN


FECN: Forward Explicit Congestion
Notification


DCE sets FECN bit to 1


When received by DTE, it indicates that
frame experienced congestion


Sent to higher layers or ignored


Congestion Control: BECN


BECN: Backward Explicit Congestion
Notification


Same as FECN but set on the return flow


LMI


Local Management Interface


Frame Relay “extension”


Introduced in 1990 by the “gang of four”
(Cisco, DEC, Nortel and Stratacom)


Additional

capabilities for complex
internetworking environments


Later Standardized by CCITT



LMI (2)


G
lobal addressing
: DLCIs become global
addresses


V
irtual
-
circuit status messages


M
ulticasting


LMI Frame Format

CIR


What you buy with a FR connection


Committed Information Rate


CIR= Committed Burst/Committed Time


Also Maximum Rate

ATM

Asynchronous Transfer Mode

Characteristics


Originally designed to transmit voice, video
and data over the same network


Cell switching


Each communication is assigned a timeslot


Timeslots are assigned on a demand
-
basis
=> asynchronous (as opposed to TDM)

Cells


53 bytes: 5 byte header + 48 byte payload


Tradeoff between voice world and data
world:


Voice needs small payloads and low delay


Data needs big payload and less overhead

ATM Interfaces


UNI: User to Network Interface


NNI: Network to Network Interface

ATM Interfaces

UNI and NNI cell formats

UNI and NNI differences


NNI has bigger VPI range


UNI has Generic Flow Control field


GFC used to identify different end stations

VPI and VCI


Used to determine paths


VPI: Virtual Path Identifier


VCI: Virtual Channel Identifier


VPI identifies a bundle of VCIs

VPI and VCI (2)

ATM Switching


Table look up


Incoming 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 model


AAL1: Connection Oriented => Voice and
Video


AAL 3,4: Connection Oriented and
Connectionless (similar to SMDS)


AAL 5: Connection Oriented and
Connectionless for CLIP and LANE


ATM Sources

ATM Addresses


ITU
-
T Standard: E.164 (Telephone #)


ATM Forum defined 20
-
byte NSAP
Addresses for use in private networks


E.164 address used as prefix on NSAP


Mapped to IP addresses by ATM ARP (in
CLIP)


ATM QoS


Traffic Contract:
peak bandwidth, average
sustained bandwidth
,
burst size
, … Similar
to FR


Traffic Shaping (end device): Queuing,
Buffering


Traffic Policing (switches): Enforces
contract

Path Establishment

LAN Emulation (LANE)


Purpose: emulate a LAN over an ATM
network


Ethernet or Token Ring


Resolves MAC addresses to ATM addresses


LANE Equivalent

LANE Components


LEC: LAN Emulation Client


LES: LAN Emulation Server


BUS: Broadcast and Unknown Server


LECS: LAN Emulation Configuration
Server

LANE Components

Initialization


LEC finds LECS via pre
-
established ILMI
procedure or through well
-
known circuit


LECS 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)


Communication


LE ARP Request sent to LES


If LES doesn’t know, it floods the request