ch04-core network qos

8/12/2019 Ch04-Core Network QoS

1/134


2/134

2Telecomm. Dept.Faculty of EEE NQoS2013HCMUT

Chapter 1: Network QoSInternet todayQoS definition,QoS parameters

Application : Voice, Video and VPNChapter 2: Per-hop QoSPacket classification: ToS, Traffic Classs, DSCP,Policing, Marking, Shaping

Queue managementSchedulingCall Admission Control (CAC)

Content


3/134


Chapter 3: Edge-to-Edge Network ModelsIntegrated Service modelDifferential Service modelEdge-to-edge IP QoSQoS and end-to-end TCP performance

Chapter 4: QoS in Core NetworksQoS in ATM networksQoS in MPLS networks

Chapter 5: QoS in Wireless NetworksQoS in WLANQoS in 3G Cellular networksQoS in Ad-hoc networks

Content


4/134


Chapter 6: Traffic EngineeringM/M/1, M/M/c modelsM/G/1, M/D/1 modelsQueuing networks

Application of queuing theory in Network QoSChapter 7: Computer simulation

TCP performance evaluation

Simulation of QoS in MPLS networksSimulation of QoS in ATM networkQoS evaluation of multimedia traffic in IP networks

Content


5/134


Rudimentary ATM Concepts

Physical layer

Signaling

Cell formatConnection types


6/134


ATM Building Blocks

ATM signalingUNI and NNI

Virtual connectionsVCC, VP, and VC


7/134


ATM Signaling

UNI = User-to-Network InterfaceNNI = Network-to-Network InterfaceCell header content varies dependingon whos talking to whom

TokenRing

Public UNI

UNI

Public ATMNetwork

NNI

NNI

NNI

Private ATM Network

Public ATMNetwork

B-ICI


8/134


Virtual Path and Virtual Channels

ATM Physical LinkVirtual Channel Connection (VCC)

Virtual Path(VP)

Contains Multiple VCs

Virtual Channel Connection(VCC)

Contains Multiple VPs

Virtual Channel(VC)

Logical PathBetween ATM End Points

Virtual Channels (VC)

Virtual Channels (VC)

E3

OC12 Virtual Path (VP)

Virtual Path (VP)

Connection Identifier = VPI/VCI


9/134


10/134

10Telecomm. Dept.Faculty of EEE

NQoS2013HCMUT

VP and VC Switching

VCI 1 VCI 2 VCI 3 VCI 4

VPI 2VPI 3VPI 1

VPI 2

VPI 3

VPI 5

VPI 1

VPI 4

Port 1

Port 2

Port 3

VCI 1

VCI 2

VCI 1

VCI 2

VP Switch

VC Switch

VCI 3

VCI 4

VCI 1

VCI 2


11/134


NQoS2013HCMUT

Virtual Channels & Virtual Paths

This hop-by-hop forwarding is known as cell relay

Virtual Channel Connection (VCC)

Virtual PathConnection (VPC)

VPSwitch

VCSwitch

VCSwitch

NNI NNI

VPI = 2VCI = 44

VPI = 1VCI = 1

VPI = 26VCI = 44

VPI = 20VCI = 30

UNIUNI


12/134


NQoS2013HCMUT


Physical layer

Signaling



13/134


NQoS2013HCMUT

Creating Cells from PacketsDest.

AddressSource

Address DataFrameCheck

PayloadHeader

Packet

Cells

PayloadHeader

PayloadHeader

PayloadHeader

SARSegmentation and Reassembly

Segmentation Happens at Source

Reassembly Happens at Destination


14/134


NQoS2013HCMUT

ATM Cell Header

5 ByteHeader

48 BytePayload

ATM Cell

53 Bytes


15/134


NQoS2013HCMUT

ATM Cell Header Details

GFC Generic Flow ControlUNI Cells Only!

VPI/VCI Identifies VirtualPaths and Channels

PTI Payload Type Identifier 3 Bits:

1. User/Control Data2. Congestion3. Last Cell

CLP Cell Loss Priority Bit

HEC Header Error Check8 Bit CRCATM NNI Cell

48 BytePayload

VPI (12) VCI (16)

PTI CLP

HEC

ATM UNI Cell

48 BytePayload

GFC (4)VPI (8)

VCI (16)

PTI CLP

HEC


16/134


NQoS2013HCMUT


Physical layer

Signaling



17/134


NQoS2013HCMUT

ATM Connection Types

PVC

SVC

Soft PVC


18/134


NQoS2013HCMUT

Connection Types

Connectionless: Packet Routing

Path 1 = S1, S2, S6, S8Path 2 = S1, S4, S7, S8

Data can take different pathand can arrive out of order

Connection Oriented: Cell Switching

VC = S1, S4, S7, S8Data takes the same pathand arrives in sequence

S2 S6

S4 S7

S3 S5

S1 S8

1

1

1

2 2

2

S2 S6

S4 S7

VC

S1 S8S3 S5


19/134


NQoS2013HCMUT

Permanent Virtual Circuit (PVC)

VPI/VCI tables in network equipment updated by administrator

A

B

D

C

Input OutputPort VPI/VCI Port VPI/VCI

1 33 3 022 15 3 14

1 64 3 29

3 29 1 64


1 29 3 452 52 4 15

1 64 3 29

3 29 1 64


1 45 2 16

2 52 1 291 64 3 29

3 29 1 64

29

52

10

16

15

4514

43


1 16 2 43

3 14 4 101 64 3 29

3 29 1 64

1

2

4 2

3

3

2

4

12

3


20/134


NQoS2013HCMUT

Switched Virtual Circuit (SVC)

Dynamically set up connections via signaling

B

D

1

2

4 1

3

3

2

4

12

UNISignaling

NNISignaling C

UNISignaling


1 64 3 29

3 29 1 64


2 52 1 291 64 3 29

3 29 1 64


1 29 3 45

1 64 3 29

3 29 1 64


1 64 3 29

3 29 1 64A

3


21/134


22/134


NQoS2013HCMUT

Switched Virtual Circuit (SVC)

Dynamically tear down connections via signaling

B

D

1

2

4 1

3

3

2

4

12

NNISignaling C

UNISignaling


1 64 3 29

3 29 1 64


2 52 1 291 64 3 29

3 29 1 64


1 29 3 45

1 64 3 29

3 29 1 64


1 64 3 29

3 29 1 64A

3

UNISignaling


23/134


NQoS2013HCMUT

Soft PVC

PVC established manually across UNI and dynamically across NNI

B

D

1

1

2

NNISignaling

UNISignaling


1 64 3 29

3 29 1 64


2 52 1 291 64 3 29

3 29 1 64


1 29 3 45

1 64 3 29

3 29 1 64


1 64 3 29

3 29 1 64

3

UNISignaling

1 29 3 45

1 16 2 43

C

A


24/134


NQoS2013HCMUT

ATM Reference model


25/134


NQoS2013HCMUT

ATM Layer

Provides VPI/VCI values in headerEnsures that cells stay in the correct order

ATMAdaptation Layer

(AAL)

ATM Layer

Virtual Channel

Connection (VCC)

Virtual PathConnection (VPC)

VPSwitch

VCSwitch

VCSwitch

NNI NNI

VPI = 12VCI = 44

VPI = 0VCI = 38

VPI = 26VCI = 44

VPI = 0VCI = 36

UNI UNI

Physical Layer


26/134


27/134

27Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

AALThe AAL segments larger packets from Frame Relay, X.25, Ethernet etc.into cells and back again. In addition it takes care of applications thatneed Constant Bit Rates (CBR) and Variable Bit Rates (VBR) . The twosublayers within the AAL that perform these functions arethe Convergence Sublayer (CS) and the Segmentation and ReassemblySublayer (SAR) . These are detailed in the adaptation layer header that sits

between the ATM cell header and the payload data.The CS provides the timing relationships between source and destinationfor CBRs and VBRs, plus it provides the correct mode for connectionoriented or connectionless services. The Common Part (CP) works withthe SAR and provides management information and the Service Specific

(SS) sublayer is specific to the type of service.The SAR examines the packets, determines the number of cells requiredfor each packet and creates SAR-PDUs which are the 48-byte payloads.The 5-byte header is then added to form the ATM cell


28/134

28Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

SAR

CS

AAL

ATM Adaptation LayerAAL

AAL = CS + SARCScell taxSARcell packet

PBXATMAdaptation Layer (AAL)

ATM Layer

Physical Layer


29/134


30/134


31/134


NQoS2013

HCMUT

AAL2


32/134


NQoS2013

HCMUT

AAL3/4


33/134

33Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

AAL5

ATM S i C t i


34/134

34Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

ATM Service Categories

Service CriteriaTraffic descriptors

QoS parameters

Service CategoriesConstant Bit Rate (CBR)

Variable Bit Rate (VBR)Unspecified Bit Rate (UBR)

Available Bit Rate (ABR)

ATM S i C it i


35/134

35Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

Contract

ATM Service Criteria

Contract

ContractTraffic Descriptors

Peak cell rate

Sustainable cell rateMaximum burst sizeMinimum Cell Rate

Quality of ServiceDelayCell loss

ATM Network

ATM S i C it i


36/134

36Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT


Peak Cell Rate PCRMaximum data ratea connection can handle without losing dataSustainable Cell Rate SCRAverage ATMcell throughput the application is permittedMaximum Burst Size MBSSize of themaximum burst of contiguous cells that

can be transmittedMinimum Cell Rate MCRRate of anapplications ability to handle latency

Traffic Descriptors



37/134

37Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT


Maximum Cell Transfer Delay MCTDHow long the network can take to transmit a cell

from one endpoint to anotherCell Delay Variation Tolerance CDVTLine distortion caused by change in interarrival

times between cells aka jitter

QoSDelay


38/134



39/134

39Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT


Service CriteriaTraffic parameters

QoS parameters

Service CategoriesConstant Bit Rate (CBR)

Variable Bit Rate (VBR)Unspecified Bit Rate (UBR)

Available Bit Rate (ABR)



40/134

40Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT


QoSTraffic Descriptor

Application

Constant Bit Rate (CBR)

Real Time Voice and Video

LOW HIGH

Tolerance

Cell DelayCell Loss

PCRPeak Cell Rate



41/134

41Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

ATM Service CategoriesVariable Bit Rate (VBR-RT/VBR-NRT)


LOW HIGH

Tolerance

Cell Delay(NRT)Cell Delay (RT)

PCR

SCR

Peak Cell Rate

Sustainable Cell Rate

MBSMaximum Burst Size Cell Loss

Packetized Voice/Video, SNA

Application



42/134

42Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT



Application

Unspecified Bit Rate (UBR)

Data Transfer

LOW HIGHNo GuaranteesSend and Pray

Tolerance

Cell LossCell Delay



43/134

43Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

ATM Service CategoriesAvailable Bit Rate (ABR)

Also usesCongestionFeedback

Mechanisms

LAN Interconnect for Data


LOW HIGHPCR

MCRPeak Cell Rate

Minimum Cell Rate

Tolerance

Cell Loss Cell Delay

Application

Traffic Management


44/134

44Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

Traffic Management

Why traffic management?Traffic control techniquesABR congestion feedback

Why Traffic Management?


45/134

45Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT


Proactively combat congestionProvision for priority control

Maintain well-behaved traffic



46/134

46Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

Ethernet (1500 Bytes) = 32 CellsFDDI (4470 Bytes) = 96 Cells

IP over ATM1577 (9180 Bytes) = 192 Cells


Lose one cell and the rest are uselessNeed to re-transmit 32+ cells for one cell lostCongestion collapse is the result

TCP/IP Packet

X

Cell Loss Datas Critical Enemy

Traffic Control Techniques


47/134

47Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT


Connection management Acceptance

Traffic management PolicingTraffic smoothing Shaping



48/134

48Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT

Contract


Contract

Contract

Traffic ParametersPeak cell rateSustainable cell rateBurst toleranceEtc.

Quality of ServiceDelay

Cell loss

ATM Network

Connection Management



49/134

49Telecomm. Dept.

Faculty of EEE

NQoS2013

HCMUT


Connection Admission Control (CAC)

ATM Network

I want a VC:X Mbps

Y DelayZ Cell Loss

CACCan I Support thisReliably without

Jeopardizing OtherContracts?

Noor

Yes, Agree to aTraffic Contract

Guaranteed QoS Request

Contract

Connection Management



50/134


NQoS2013HCMUT


ATM Network

You areNot in Conformance

with the Contract.What Should the

Penalty Be??

PASS MARK CLP BIT DROP

?DECISION?

Contract

Usage Parameter Control (UPC) aka Policing

REBELAPPLICATION

Traffic Management



51/134


NQoS2013HCMUT


CLP ControlWhen congested, drop marked cellsPublic UNIGeneric Cell Rate Algorithm (GCRA)

0 0 0 0 1 0Marked

UPC

PASS MARK CLP BIT DROP

?DECISION?

Traffic Management

Dr op



52/134


NQoS2013HCMUT

q

Tail Packet Discard (TPD)Discard cells from same bad packet

0 0 0 0 1 0Marked

UPC

Traffic Management

3 2

Dr op



53/134


NQoS2013HCMUT

ATM/Switch

6

1 0

5

4

Output Buffer EOM

UPC

7 3 2 1

X

Intelligent Tail Packet Discard

Output Queue

EOM

EPD Threshold

EOM

Early Packet Discard

aka UBR+

qTraffic Management



54/134


NQoS2013HCMUT

Switch without Packet Discard

Switch with Intelligent Packet Discard

Maximize Goodput

qTraffic Management


55/134



56/134


NQoS2013HCMUT

RMResource Management cells

Rate-based feedback mechanisms:EFCI markingRelative rate marking

Explicit rate markingVS/VD

ABR Congestion Feedback


57/134



58/134


NQoS2013HCMUT

Relative rate markingSwitches can set congestion flagin backward RM cells

Source

Forward

BackwardCongestionExperiencedSlow Down

RM X

RM


Destination



59/134


NQoS2013HCMUT


Explicit rate markingSwitches can tell source at exactly what rate to transmit

Source

Destination

Forward

Backward

RM

RM

Congestion ExperiencedSlow Down X Amount

CongestionExperienced

Slow Down



60/134


NQoS2013HCMUT

VS/VDVirtual source/virtual destinationBreaks the feedback loop into separate segmentsShortens length of feedback loop

Source

Destination

Forward

Backward

CongestionExperienced

Slow Down




61/134


NQoS2013HCMUT

Absorb traffic bursts from simultaneous connectionsSwitches schedule traffic based on priorityof traffic according to QoSSwitch must reallocate buffers as the traffic mixchangesEffective buffering maximizes throughput

of usable cells as opposed to raw cells(aka goodput)

Buffers Are Your Friend

IP/ATM Model


62/134


NQoS2013HCMUT

Mapping cells into SONET/SDH

Using logical IP subnet (LIS) instead of using regular IPsubnetTwo models: Classical IP/ATM (CIP) and LAN Emulation(LANE) for IP/ATMClassical IP/ATM (CIP): link driver has two concerns:

Address resolution: discovering the link-level next-hopcorresponding to a given IP next-hop address known to be onthe linkPacket Transport: actually establishing link-level resources totransfer the packet to the indicated next-hop

AAL5 is used for encapsulation of IP packets For GS use CBR or rt-VBR For CL use non-rt VBR or ABR with a minimum cell rate For BE use UBR or ABR

IP/ATM QoS


63/134


NQoS2013HCMUT

Controlling QoS in IP/ATM involves 2 aspects:Upgrading the IP node with sufficient CQS architecture toprovide differentiated scheduling of traffic toward any next-hopDefining rules for establishing and using parallel ATM VCstoward a given next-hop


64/134

Internet core network


65/134


NQoS2013HCMUT

Router based core network

Internet core network (cont)


66/134


NQoS2013HCMUT

Processing can not meet bandwidth demands

Bottle-neck in software-based routersAvailable router interfaces not provide trafficaggregation

Metric-based routing was no longer scalableDensely connected networks lead to inefficient use ofnetwork resourcesDestination based routing tends to aggregate alltraffic to the same destination: not utilize links

Internet core network (1)


67/134


NQoS2013HCMUT

Switch based core network



68/134


NQoS2013HCMUT

Faster and simpler forwarding, better traffic aggregation

Fix size cells can be handled in hardware to speed upConnection-oriented forwarding algorithm improves

performance gain: based on short fix length connectionidentifiers

The ASIC technology : IP packets can be forwarded with highspeed. ATM interfaces have even fallen behind the latestincreases of optical network (packet over SDH/SONET)

Waste of bandwidth : 5/48 bytes of header.

Complex network management: physical ATM switchedinfrastructure and logical IP network topology. Each layer usesits own addressing scheme and routing protocol



69/134


NQoS2013HCMUT

The n-squared problem : when adding or shutting

down any router will create enormous signaling loadIGP stress : intra-domain routing is not conceived for

fully meshed topology. With high number of routingpeer routers: too much routing information has to beexchanged

Multi-Protocol Label Switching (MPLS) can offer

solutions that create a combination of the advantagesfrom both of these worlds

MPLS NGN


70/134


NQoS2013HCMUT

MPLS NGN (cont)


71/134


NQoS2013HCMUT

Forwarding based on label: speed up processing at node

Forwarding mechanism: label (explicit routing) or IP header(hop-by-hop routing)

Operating on any layer 2 technologies: ATM, Ethernet, FR

Allows for both traffic aggregation and disaggregation

Support VPN : using 64-bit VPN address (total 92 bits)Allow SPs embed into the IP network : TE and traditional QoS of

layer 2 : using DSCP and processing queues based on its packetspriority

Easy management and operation

Traditional IP Routing


72/134


NQoS2013HCMUT

Choosing the next hop

Open Shortest Path First (OSPF) to populate the routingtable

Route look up based on the IP addressFind the next router to which the packet has to be sent

Replace the layer 2 address

Each router performs these steps


73/134


74/134

Distributing Routing Information(1)


75/134


NQoS2013HCMUT

3

1

0125.50

145.40

2

AddressPrefix

Address

Prefix

AddressPrefix

Path Path

Path

125.50

145.40

125.50

145.40

125.50

13

3 0

2

Data 125.50.33.85Data 125.50.33.85

Disadvantages


76/134


NQoS2013HCMUT

Header analysis performed at each hop

Increased demand on routersUtilizes the best available pathSome congested links and some underutilized links!

Degradation of throughputLong delays

More losses

No QoSNo service differentiation

Not possible with connectionless protocols

Need for MPLSR id th f I t t


77/134


NQoS2013HCMUT

Rapid growth of Internet

New latency dependent applicationsQuality of Service (QoS)

Less time at the routers

Traffic EngineeringFlexibility in routing packets

Connection-oriented forwarding techniques withconnectionless IP

Utilizes the IP header information to maintain interoperability with IP

based networksDecides on the path of a packet before sending it

What is MPLS?l l l h h


78/134


NQoS2013HCMUT

Multi Protocol supports protocols even other than IP

Supports IPv4, IPv6, IPX, AppleTalk at the network layerSupports Ethernet, Token Ring, FDDI, ATM, Frame Relay, PPPat the link layer

Label short fixed length identifier to determine a

routeLabels are added to the top of the IP packetLabels are assigned when the packet enters the MPLSdomain

Switching forwarding a packetPackets are forwarded based on the label valueNOT on the basis of IP header information

MPLS BackgroundIntegration of layer 2 and layer 3


79/134


NQoS2013HCMUT

Integration of layer 2 and layer 3Simplified connection-oriented forwarding of layer 2

Flexibility and scalability of layer 3 routing

MPLS does not replace IP; it supplements IP

Traffic can be marked, classified and explicitly routed

QoS can be achieved through MPLS

IP/MPLS comparison


80/134


NQoS2013HCMUT

Routing decisions

IP routing based on destination IP addressLabel switching based on labels

Entire IP header analysis

IP routing performed at each hop of the packets path in the network

Label switching performed only at the ingress router

Support for unicast and multicast data

IP routing requires special multicast routing and forwarding

algorithmsLabel switching requires only one forwarding algorithm


81/134

Forwarding Equivalence Class (FEC)A f k h i h f di


82/134


NQoS2013HCMUT

A group of packets that require the same forwarding

treatment across the same pathPackets are grouped based on any of the following

Address prefixHost addressQuality of Service (QoS)

FEC is encoded as the label


83/134

FEC example (contd)Assume packets have the destination address and QoS


84/134


NQoS2013HCMUT

Assume packets have the destination address and QoSrequirements as124.48.45.20 qos = 1143.67.25.77 qos = 1143.67.84.22 qos = 3124.48.66.90 qos = 4143.67.12.01 qos = 3

FEC 1 label a FEC 2 label b FEC 3 label c FEC 4 label d143.67.25.77 124.48.45.20 143.67.84.22 124.48.66.90

143.67.12.01

Example of a MPLS network


85/134


NQoS2013HCMUT


86/134

Label Switching Router (LSR)A router/switch that supports MPLS


87/134


NQoS2013HCMUT

A router/switch that supports MPLS

Can be a routerCan be an ATM switch + label switch controller

Label swapping

Each LSR examines the label on top of the stackUses the Label Information Base (LIB) to decide theoutgoing path and the outgoing label

Removes the old label and attaches the new label

Forwards the packet on the predetermined path

Label Switched Path(LSP)LSP defines the path through LSRs from ingress to


88/134


NQoS2013HCMUT

LSP defines the path through LSRs from ingress to

egress routerFEC is determined at the LER-ingressLSPs are unidirectional

LSP might deviate from the IGP shortest path

LSP

LSP

LDP


89/134


NQoS2013HCMUT

Label Distribution Protocol

LDP


90/134


NQoS2013HCMUT

LDP


91/134


NQoS2013HCMUT

LDP - AdvantagesExplicit routing


92/134


NQoS2013HCMUT

p g

Set up a LSP between Ingress Router and EgressRouterLabel request for each hop on down-streamLabel mapping : up-streamErrors occur: router sends a alarm message toneighbors or operating routers to re-direct for currentLSP

Less resources (compared with RSVP)

LDP - DisadvantagesSlow error recovery


93/134


NQoS2013HCMUT

y

Not support dynamically re-optimization of trafficflowsTransient periods: efficiency of Resource Location

could be influenced by routing traffic.Require means to restore the LSP to the original

routes once congestion has subsidedFATE : using dynamic reroute mechanism

Shim HeaderA short, fixed length identifier (32 bits)


94/134


NQoS2013HCMUT

, g ( )

Sent with each packetLocal between two routersCan have different labels if entering from different routersOne label for one FEC

Decided by the downstream routerLSR binds a label to an FECIt then informs the upstream LSR of the binding


95/134

Time To Live (TTL)TTL value decremented by 1 when it passes through an


96/134


NQoS2013HCMUT

y p g

LSRIf TTL value = 0 before the destination, discard thepacket

Avoids loops may exist because of somemisconfigurations

Multicast scoping limit the scope of a packetSupporting the traceroute command

TTL (contd)Shim header


97/134


NQoS2013HCMUT

Has an explicit TTL fieldInitially loaded from the IP header TTL fieldAt the egress LER, value of TTL is copied into the TTL field ofthe IP header

Data link layer header (e.g VPI/VCI)No explicit TTL fieldIngress LER estimates the LSP lengthDecrements the TTL count by the LSP lengthIf initial count of TTL less than the LSP length, discard the

packet


98/134


99/134


100/134

Label stack (contd)


101/134


NQoS2013HCMUT

Labels scope and uniquenessLabels are local between two LSRs


102/134


NQoS2013HCMUT

Rd might give label L1 for FEC F and distribute it to Ru1At the same time, it might give a label L2 to FEC F anddistribute it to Ru2L1 might not necessarily be equal to L2

Can there be a same label for different FECs?Generally, NOBUT no such specificationLSR must have different label spaces to accommodate bothSHIM header specifies that different label spaces used forunicast packets and multicast packets

Invalid labelsWhat should be done if an LSR receives an invalidl b l?


103/134


NQoS2013HCMUT

label?Should it be forwarded as an unlabeled IP packet?

Should it be discarded?

MUST be discarded!

Forwarding it can cause a loopSame treatment if there is no valid outgoing label


104/134

Route selection (contd)Hop by Hop


105/134


NQoS2013HCMUT

Allows each LSR to individually choose the next hopThis is the usual mode today in existing IP networks

No overhead processing as compared to IP

Explicit routingA single router, generally the ingress LER, specifies severalor all of the LSRs in the LSP

Provides functionality for traffic engineering and QoSo Several: loosely explicitly routed

o All: strictly explicitly routed

E.g. CR-LDP, TE-RSVP

Label Information Base (LIB)Table maintained by the LSRs


106/134


NQoS2013HCMUT

Contents of the tableIncoming label

Outgoing labelOutgoing path

Address prefixIncominglabel Address Prefix

OutgoingPath

Outgoinglabel

LSR Forwarding Engine


107/134


NQoS2013HCMUT

MPLS forwardingExisting routing protocols establish routes


108/134


NQoS2013HCMUT

LDP establishes label to route mappingsLDP creates LIB entries for each LSR

Ingress LER receives packet,adds a labelLSRs forward labeled packets using label swapping

Egress LER removes the label and delivers the packet

FEC in MPLS


109/134


NQoS2013HCMUT

MPLS forwarding (contd)

AddressPrefix

OutPath

InLabel

OutLabel


110/134


NQoS2013HCMUT

Use label 8 for 145.40

Use label 9 for 125.50

Use label 2 for 125.50 andlabel 1 for 145.40

3

1

0125.50

145.40

2

AddressPrefix

OutPath

InLabel

OutPath

InLabel

OutLabel

OutLabel

125.50 125.50

125.50

145.40 145.40

3

3

2

1

2

1 8

9

9

0

1

2Address

Prefix

MPLS forwarding (contd)

AddressPrefix

OutPath

InLabel

OutLabel


111/134


NQoS2013HCMUT

3

1

0125.50

145.40

2

AddressPrefix AddressPrefixOutPath InLabel OutPath InLabelOutLabel OutLabel

125.50 125.50

125.50

145.40 145.40

3

3

2

1

2

1 8

9

9

0

1

2

Data 125.50.33.85 2

Data 125.50.33.85 9


112/134


NQoS2013HCMUT

DiffServ & MPLS

112

DiffServ Architecture


113/134


NQoS2013HCMUT

Differentiated ServicesThe IETF DiffServ Model


114/134


NQoS2013HCMUT

Use 6 bits in IP header to sort traffic intoBehavior AggregatesAKA Classes!Defines a number of Per Hop Behaviors - PHBsTwo-Ingredient Recipe:

Condition the Traffic at the Edges Invoke the PHBs in the Core

Use PHBs to Construct Services such as VirtualLeased Line!

Defined PHBs

Expedited Forwarding (EF): RFC2598


115/134


NQoS2013HCMUT

p g ( )dedicated low delay queueComparable to Guaranteed B/W in IntServ

Assured Forwarding (AF): RFC2597

4 queues 3 drop preferencesComparable to Controlled Load in IntServClass Selector: Compat. with IP PrecDefault (best effort)

AF PHB Group DefinitionAF Class 1: 001 dd 0


116/134


NQoS2013HCMUT

4 independently-forwarded AF classesWithin each AF class, 3 levels of drop priority! This is very useful to protectconforming to a purchased, guarantee rate, while increasing chances ofpackets exceeding contracted rate being dropped if congestion isexperienced in the core.

AF Class 2: 010 dd 0



Eg. AF12 = Class 1, Drop 2, thus 001100

dd = drop preference

DiffServ Scalability via Aggregation

1000sof flows


117/134


NQoS2013HCMUT

DiffServ scalability comes from:- aggregation of traffic on Edge- processing of Aggregate only in Core

Diff-Serv:

Aggregated Processing inCore

Scheduling/Dropping (PHB)based on DSCP

Diff-Serv:


Scheduling/Dropping (PHB)based on DSCP

Diff-Serv:

Aggregation on Edge

Many flows associated witha Class (marked with DSCP)

Diff-Serv:

Aggregation on Edge

Many flows associated witha Class (marked with DSCP)

MPLS Scalability via Aggregation

1000sof flows


118/134


NQoS2013HCMUT

MPLS scalability comes from:- aggregation of traffic on Edge- processing of Aggregate only in Core

MPLS:


Forwarding based on label

MPLS:


Forwarding based on label

MPLS:

Aggregation on Edge

Many flows associated witha Forwarding EquivalentClass (marked with label)

MPLS:

Aggregation on Edge

Many flows associated witha Forwarding EquivalentClass (marked with label)

MPLS & DiffServ - The Perfect Match!

1000sof flows


119/134


NQoS2013HCMUT

Because of same scalability goals, both models do:- aggregation of traffic on Edge- processing of Aggregate only in Core

MPLS: flowsassociated withFEC, mapped

into one label

MPLS: flowsassociated withFEC, mapped

into one label MPLS:Switchingbased onLabel

MPLS:Switchingbased onLabel

DS:Scheduling/Droppingbased on DSCP

DS:Scheduling/Droppingbased on DSCP

DS: flows associatedwith Class, mappedto DSCP

DS: flows associatedwith Class, mappedto DSCP

Non-MPLSDiff-Serv Domain

MPLSDiff-Serv Domain

MPLS - The Shim Header!!


120/134


NQoS2013HCMUT

DSCPfield is not directly visible to MPLS Label Switch Routers (theyforward based on MPLS Header)

Information on DiffServ must be made visible to LSR in MPLS Header(using EXP field / Label)

0 1 2 30 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

| Label | EXP |S| TTL |

DSCP

IPv4 Packet MPLS Header

DSCP

Driving PHB at Core LSR


121/134


NQoS2013HCMUT

Driving PHB at Core LSRQoS consideration

Each distinct queuing and


122/134


NQoS2013HCMUT

scheduling behavior may beencoded as a new FEC (LSP),ignoring the EXP fieldThe EXP field encodes up to 8

queuing and schedulingbehaviors for the same FEC (LSP)

The EXP field encodes up to 8queuing and scheduling

behaviors independent of FEC(LSP)

Driving PHB at Core LSRSome possible approaches:

Using Label to select a queue (Service Class) and using one or


123/134


NQoS2013HCMUT

more EXP bits to encode different drop precedence levelsUsing Label to select a group of four queues, using 2 bitsfrom EXP to select one of those 4 queues and using theremaining bit from EXP to encode the drop precedence

Ignoring the Label and using four shared queues per outputinterface. 2 bits from EXP to select one of those 4 queues,and the remaining bit from EXP encodes the dropprecedence

Using Label to select a group of N queues (N


124/134


NQoS2013HCMUT


125/134


NQoS2013HCMUT

MPLS Traffic Engineering

125

MPLS TE & QoS The RelationshipMPLS TE designed as tool to improve backboneefficiency independently of core QoS


126/134


NQoS2013HCMUT

techniques:MPLS TE compute routes for aggregates across all PHBs. A Single Chunk of Bandwidth requested for the Tunnel

MPLS TE performs admission control over a global b/w pool. Un-aware of bandwidth allocated to each Class / PHB

MPLS TE and MPLS DiffServ:Can run simultaneously in a network.Can provide their own individual benefits TE distributes aggregate load DiffServ provides differentiation

Are unaware of each other

Traffic Aggregate in IP networks


127/134


NQoS2013HCMUT

Traffic Aggregate in MPLS networks


128/134


NQoS2013HCMUT

Traffic Aggregate in MPLS networks


129/134


NQoS2013HCMUT

TE Fast Reroute - Tunneling


130/134


NQoS2013HCMUT

TE Global Fast Reroute (Makam)


131/134


NQoS2013HCMUT

TE Region Fast Reroute


132/134


NQoS2013HCMUT

TE Local Fast Reroute


133/134


NQoS2013HCMUT

TE Haskin Fast Reroute


134/134


NQoS2013HCMUT

ch04-core network qos

Documents