transport networks and transport mpls - intranet...

Transport Networks andTransport MPLS

Alberto Lometti

March, 2007

2 | Transport MPLS Rationale and Features | February 2007 All Rights Reserved © Alcatel-Lucent 2006, 21407

Agenda

1. Transport network basics

2. TMPLS rationale

3. TMPLS features

4. TMPLS standard status & perspectives

5. A multi-layer multi-technology transport network

6. Appendix: basics of GMPLS CP protocols


1 Transport network basics


“E2E” transport network

Operator A

Service provider

ClientClient Operator n

“E2E” transport network

Service

ServiceNetworkNode

Service “transparent” transport


Transport service characteristics

By means of mid-long termcountermeasures againstfailures(~h ÷ ~d)

In terms of service quality,according to the differenttechnology type

By means of short termcountermeasures againstfailures(50 msec ÷ ~sec)

Maintainable

Measurable

ResilientAvailable (% UAT)

For the client For the operator


A simplified model for transport networks

Client A Client AC

lient

B

Clie

nt B

Clie

nt C

Clie

nt C

Client, requiringtransportservice

UNI, interfacebetween the client

and the serviceprovider

Logical connection, related to thetransport service and characterized bybeing continuous along the service

path within the network(end-to-end)

Physicalconnection,

i.e. thephysicaltransportmedium

Connection matrix,operating on logical

connection andallowing for service

grooming and routingMore than one layer is possible (and typically isused) at logical connection level. Each layer, is

independently operated from the other ones

NNI,interfacebetween

two networknodes


Transport service functional model

Client

Client

Non intrusive monitoring:• integrity check• connectivity check• quality check

Logicalconnection

Client-server adaptation• forward defect indication

Termination:Like non intr. mon. +• backward integrity info.• backward quality info.

Connection:• forwarding• grooming

Physicalconnections

Client-server adaptation• forward defect indication

Termination:Like non intr. mon. +• backward integrity info.• backward quality info.


Fiber

ConnectionConnection

Integ rityInteg rity

Remore qual.Remore qual.

Remore qual.

QualityQuality


QualityQuality


QualityQuality

Integ rityInteg rityQualitymeasurement ConnectionConnection

QualityQuality

Remote integ r.

Integ ritySilencingInteg rityInteg ritySilencing

Maintenance

Cong ruenceCong ruenceCong ruence

ConnectionConnectionConnection

QualityQualityQuality

Remote integ r.Remote integ r.

Remore qual.

ConnectionConnectionProtection

Remote integ r.


ClientLogical connection

Physical connection

Simplified network model & basic OAM tools


Fiber




Remore qual.

QualityQuality


QualityQuality


QualityQuality


QualityQuality

Remote integ r.


Maintenance

Cong ruenceCong ruenceCong ruence




Remore qual.

ConnectionConnectionProtection

Remote integ r.



Physical connection

Basic OAM tools usage: an example (1/3)

Service protection is typically based onphysical line (because it is the one that breaks)and logical connection (because it is related to

the end-to-end service within the network).


Fiber




Remore qual.

QualityQuality


QualityQuality


QualityQuality


QualityQuality

Remote integ r.


Maintenance




Remore qual.


Protection

Remote integ r.



Physical connection


Service maintenance is typically based on physical linecheck and can be helped by logical connection and client

signals inspection; in addition downstream alarm silencing isperformed


Fiber




Remore qual.

QualityQuality


QualityQuality


QualityQuality

Integ rityInteg rityQuality

measurement ConnectionConnection

QualityQuality

Remote integ r.


Maintenance




Remore qual.


Protection

Remote integ r.



Physical connection


Quality of service is typically measured on the logicalconnection both at termination and non-intrusive

monitoring points; possibly also at client non-intrusivemonitoring points. Remote indications allow for

bidirectional service monitoring making enquiries to asingle system.


SDH according to simplified model

Client A Client AC

lient

B

Clie

nt B

Clie

nt C

Clie

nt C

Client: e.g.PDH, Ethernet

Logical connection:VC12, VC3, VC4

(Virtual Container)

Physical connection:STM-N

(from 155Mb/s to 10Gb/s)

Connectionmatrices operate

at VC4 and/orVC12,VC3 level


SDH (SONET) OAM – as an example

Fiber

TTId Mism .Client specif ic

AU (TU) SFClient specif ic

MS_ REIREI

MS_ REI

EXBER, DEGEXBER, DEG

AU (TU) SFClient specif ic

BIP-NClient specif ic

TTId Mism .Client specif ic

EXBER, DEGClient specif ic

AU (TU) SFClient specif icQualitymeasurement TTId Mism .Client specif ic

BIP-NClient specif ic

RDI

LOS, LOFAU (TU) AISAU (TU) SFAU (TU) SFClient AIS

Maintenance

TTId Mism .TTId Mism .TTId Mism .

EXBER, DEGEXBER, DEGEXBER, DEG

MS-RDIRDI

REI

TTId Mism .TTId Mism .Protection

MS-RDI

LOS, LOFAU (TU) SF


Physical connection


OTN according to simplified model

Client AC

lient

B

Clie

nt B

Clie

nt C

Clie

nt C

Client: e.g. SDH(STM16, STM64)or Ethernet (GE,

10GE)

Logical connection:ODU-N (N=1, 2, 3 for λ

at 2,5Gb/s, 10Gb/s,40Gb/s)

Physical connection: OTM-x, (x λWDM signal); OTU-N (N=1, 2, 3 for λ

at 2,5Gb/s, 10Gb/s, 40Gb/s)

Connectionmatrices operateat ODU1, ODU2,

(ODU3) level

Client A


OTN OAM – as an example

ODU TTId Mis.Client specif ic

OTU SFClient specif ic

OTU_ BEIODU-BEI meas.

OTU_ BEI

EXBER, DEGDEG

OTU SFClient specif ic

BER meas.Client specif ic

ODU TTId Mis.Client specif ic

EXBER, DEGClient specif ic

OTU SFClient specif icQuality

measurement ODU TTId Mis.Client specif ic

BER meas.Client specif ic

ODU-BDI

LOS, LOFAU (TU) AISOTU SFOTU SFClient AIS

Maintenance

OTU TTId Mis.ODU TTId Mis.ODU TTId Mis.

EXBER, DEGDEGDEG

OTU-BDIODU-BDI

ODU-BEI meas.

OTU TTId Mis.ODU TTId Mis.Protection

OTU-BDI

LOS, LOFOTU SF

Fiber


Physical connection


2 T-MPLS Rationale


TDM/WDM-based

networking

Packet/cell/frame basednetworking

Reliability, availability,OAM in service delivery

Scalability and granularity

Service delivery throughpacket-based technologies

+ PacketTransport Networking

Transport network evolution (min cost) with

(max carrier-class& flexibility)

Services

Service Evolution

Networking Requirements

Why a Packet Transport Network?

Triple Play Business Services

Migrating to data-orientedleased lines, virtual private LANs

Mobile Service Operators

Converging transport of radio acc., pt-to-pt mw, NGN services…

Driving withHSI, video, voice


Packet Transport Network: Requirements

Being client-agnostic (able to carry L3, L2, L1)

Being strictly connection-oriented

Implementing strong OAM capabilities, similar to those available in existingtransport networks (e.g., SDH)

Implementing both service and line resilience mechanisms, similar to thoseavailable in existing transport networks (e.g., SDH)

Allowing for network provisioning either via a centralized management tooland/or a distributed CP

Allowing for homogeneous and/or unified management and control oftransport network when different layers are present at the same time

These requirements are the basis of ITU T-MPLS design.


Expanding the Transport Network Architecture with Packets

Management Plane

Control Plane

Data Plane

Survivability (Protection, Restoration)

OAM

SDH LayersFraming, Forwarding,

Encapsulation

OTN/WDM LayersFraming, Forwarding,

Encapsulation

T-MPLS LayersFraming, Forwarding,

Encapsulation

Common multilayer operations, survivability, control and management paradigms forpackets, circuits & photonics


3 T-MPLS features


Transport MPLS (T-MPLS) in Brief

Transport MPLS is a connection-oriented packet transport technology,based on MPLS frame formats.

It reuses the most widespread label swapping paradigm existing in telecom.

It inherits all IETF definition activity in terms of server and client encapsulationrules.

It profiles MPLS so that it avoids the complexity and need for IP routingcapability.

Transport MPLS defines powerful OAM capabilities that enable status &performance reports, in such a way that they remain confined within the T-MPLSlayer and do not require deeper packet inspection.

It allows for guaranteed SLAs.

It defines protection switching and restoration.

It allows for efficient fault localization and multi-operator service offering.

A Transport MPLS layer network is operated with network management and/or bya control plane.

The control plane inside the T-MPLS network is GMPLS.


Client D

Client A

Clie

nt C

Clie

nt B

Clie

nt C

T-MPLS network model

Clie

nt B

Client A

Client D

Clients: in T-MPLS it canbe any L2, using mappingoptions defined by IETF

for PWE3: Ethernet, ATM,FR … ) or IP/MPLS (so faronly Ethernet formally

defined in ITU)

Physical connection: inT-MPLS it can be any L1

(Ethernet, SDH, OTN+WDM…); moreover in T-MPLS an

optional section layer isunder definition, in order

to support OAM if nototherwise available

Logical connection: inTMPLS two networking

layers are underdefinition: circuit

(equivalent to IETF PWE3)and path (equivalent to

IETF MPLS tunnel)

Logical connection: inT-MPLS two networking

layers are underdefinition: circuit

(equivalent to IETF PWE3)and path (equivalent to

IETF MPLS tunnel)

T-MPLS equipment(switching capability either

at circuit or path layer)


Packet Transport Network “ideal” Modeling

Clientservices

Networkservicelayer(s)

mp2mpadaptation

p2padaptation

Logical transport(1:1 with service)

Transportlayer(s)

Logical transport(N:1 with service)

E2e & segment OAME2e & segment protectionSLA enforcement

Trunk OAM,protection & restoration(combined)Service aggregation

Recursivetransportlayer(s)

L0-L1networking

Physicallayer

EthernetSDH, OTH

“Ideal” model

Any L1 (via CES), L2, L3 service

Mp2mp and p2p adaptations may not bepart, in strict sense, of the packet transporttechnology. As an example, in VPLS mp2mpadaptation is based on Ethernet.

Two true networking layers are (at least)typically needed for a transport solution.The first layer is the carrier devoted to aspecific service, and thus allow for e2eOAM, performance monitoring andprotection. It needs not to be associatedend to end with the second layer, fornetwork scalability purposes. It is the basictool for SLA enforcement. The second layerallows mainly for aggregation and thusscalability. Restoration is possibly appliedto this layer. Of course further aggregationlayers can be recursively defined. Ingeneral, OAM and protection are data planemechanisms, generally implemented in HWfor fast and deterministic behavior.


MPLSprotocolssuite asdistributedcontrol plane

IETF MPLS:• A complete toolset for

L1/L2/L3 servicemanagement, withtransport capabilities

• IP-centric and serviceoptimized

• Coherence amongMPLS Service andTransport Layers

Packet Transport Network Modeling & MPLS

Clientservices


mp2mpadaptation


Transportlayer(s)





L0-L1networking

Physicallayer

EthernetSDH, OTH

MPLS“Ideal” model


MS-PW (understandardization)

MPLS PSN

VCCV

BFD, PingFRR

VPLSIP-VPN

PWVCCV

SDHOTH Ethernet

SDHOTH

p2padaptation


ITU-T T-MPLS:• A packet technology

optimized fortransport purposes

• Carrier focused• Coherence among

packet and TDM/λtransport

• MPLS used as servicelayer only

Packet Transport Network Modeling & TMPLS

Clientservices


mp2mpadaptation


Transportlayer(s)





L0-L1networking

EthernetSDH, OTH

TMPLS“Ideal” model


CentralizedNM orASON/GMPLSprotocolssuite asdistributedcontrol plane

TMPLS path

TMPLS circuitY.17tomG.8131 (linearprotections)

Y.17tomG.8131 (linearprotections)G.8132 (ringprotection)

VPLSIP-VPN

PW

Eth.only

L1 (CES)Any L2

SDHOTH Ethernet

SDHOTH

p2padaptation


Packet Transport Network Modeling: MPLS-TMPLS synergies

Clientservices


mp2mpadaptation


Transportlayer(s)





L0-L1networking

Physicallayer

EthernetSDH, OTH

MPLS

VPLSIP-VPN

PW

MS-PW (understandardization)

MPLS PSN

TMPLS

TMPLS path

TMPLS circuit

“Ideal” model


TMPLS service architecture is inherited byMPLS. TMPLS enhances MPLS thanks to itstransport capabilities, in line with andoperationally similar to SDH and OTH.GMPLS CP common to packets, TDM andlambdas opens the possibility of multi-layernetwork optimizations. Identical data planesubset and encapsulation rules allow veryefficient MPLS-TMPLS inter-workingschemes. ITU “tandem connection” conceptideally complements IETF OAM capabilities.

SDHOTH Ethernet

SDHOTH

SDHOTH Ethernet

SDHOTH

p2padaptation


T-MPLS Survivability

Protection (data plane) <50 ms with automatic protection switch

(APS) protocols on data plane OAM channels 1+1, 1:1, N:1 (no extra traffic) Unidirectional, bidirectional Section, path, circuit Subnetwork connection (SNCP) Ring Dual node interconnect (DNI)

T-MPLS

ClientEquipment

Restoration (control/management plane) Distributed control plane (ASON/GMPLS) Full LSP rerouting restoration Pre-planned LSP rerouting restoration Any topology (mesh) GMPLS-based restoration in synergy with

other transport network technologies (SDH,OTN, WDM)

RingProtect

DNI

Circuit & PathProtection

Section Protection

ClientEquipment


T-MPLS OAM

T-MPLS

Y.1711 - Y.17tomOAM primitives

Not applicable

Loopback

Frame Loss, Frame

Delay, FDV

Link Trace

(ffs)

Loopback

On demand

FDISilencing

BDIRemote integrity

PM (carried by FFD)Quality

FFDConnectivity

FFD, CVIntegrity

Continuous

Standardized

Ongoing


T-MPLS OAM according to ITU is contained in MPLS frames associated to the connection by therelevant LSP (outer) label and characterized by an (inner) label 14. OAM information is

contained within the MPLS payload.

Fiber (λi)

Client

Logical Connection

Physical Connection

QualityMeasurement

Maintenance

Protection

Link TraceClient specific

CV, FFDClient specific

Server specificRemote quality

Server specific

Server specificPM

CV, FFD,Client specific

PMClient specific


PMClient specific

CV, FFD,Client specific


PMClient specific

BDI

Server specificFDICV, FFDCV, FFDSilencing

Server specificCV, FFDCV, FFD

Server specificPMPM

Server specificBDI

Remote quality

Server specificFFD (via TTSId)

Server specificBDI

Server specificFFD

T-MPLS OAM(according to ITU-T Y.1711, Y.17tom)


T-MPLS Control Plane (ASON/GMPLS)

GMPLS is a single, generalized distributed control plane that can be used incommon for multiple networking technologies, including packets, TDM andphotonicsGMPLS defines: UNI concept (thus easing overlay dynamic approach) I(nternal)-NNI concept (the only supported in MPLS CP) E(xternal)-NNI concept (thus allowing for interworking among different

vendors/operators) Bidirectional pathsGMPLS allows for separation of data plane and control plane Only control interfaces are used to flood control informationGMPLS allows for “horizontal” scalability in routing domains (thanks toseparation of data plane and control plane and recursive topology)GMPLS allows for “vertical” scalability (same control plane across differentlayers)

GMPLS is the ideal control plane for multilayered networks


4 T-MPLS Standard Status & Perspectives


Protection G.8131

OAM Y.17tom

GMPLS ProtocolsControl Plane &

Restoration(IETF)

T-MPLS Standards Framework

MPLS Architecture/

Protocols(IETF)

Carrier-grade packet forwarding best from IETF

Carrier-grade control planebest from IETF

TMPLSFramework

Contribute

T-MPLS InterfacesG.8112

T-MPLS EquipmentFunctional Blocks

G.8121

Architecture of T-MPLS Layer

Network G.8110.1

ASON ArchitectureG.8080

Best from ITU


Transport Network Standard Overview

T-MPLSstandard status

Amend. 1 (ring) to beconsented Jun07

Y.1720 approvedG.8131 v1 (adaptationto T-MPLS) consented

Version 1 approvedAmend. 1 (p2mp,protection) to beconsented Mar-Jun07

Y.1711 approvedY.17tom (ext. OAM, TC)to be consented Apr07

Version 1 approvedAmend. 1 (p2mp) to beconsented Mar-Jun07

Version 1 approvedAmend. 1 (p2mp,extended OAM) to beconsented Mar-Jun07

SDH OTH T-MPLSC.O. G.805C.L. G.809

G.873.2 (frozen)

G.808.2(frozen)

Y.17tom (Y.1711)

G.8121 (G.mplseq) G.806

G.808.1

----

G.8131 (Y.1720)

G.803 G.872 G.8110.1

G.841

G.709

G.8132Ring Protection

Linear Protection G.873.1

Network Architecture

Equipment G.783 G.798

UNI & NNI Interface

OAM [frame format]

G.707

Connection Oriented(G.805) Generic

C.O. + C.L. G.uftan

G.8112 (G.motnni) ----


5 A multi-layer multi-technologytransport network


Data-centric network abstraction

ADMADM

ADMADM

Layer 1 (TDM and/or λ)

XC

XC

XCXC

Layer 2CE

L2(L3) businessservices

Residential services

PE

Layer 3

Metro aggregation CoreAccess

ADMADM

ADMADM

Mobile services

CPE (res.)

CPE (bus.) Aggregationtier 1

Aggregationtier 2

Edge

Outer Core Inner Core

BB access ADMADM ADMADMXC XC

Optical AccessCPE (bus.)


Key transport networking needs

•Ethernet demarcation at(business) user-network interface

•Circuit emulation capabilitiesin order to cross pure packetbased access areas with legacy,TDM based, services

•Multi-service encapsulationcapability, for convergingdifferent types of L2 services(ATM, FR, Ethernet …) on acommon packet infra-structure inview of coherent end-to-endservice management

•Optical access solutions forFTTx, either point-to-point orWDM PON based

Access

CPE (res.)

•Multi-service aggregationcapability, for converging newbusiness and residential packetbased with legacy ones along with2G and 3G mobile backhauling

•Strong OAM capabilities, andefficient (linear & ring)protections to manage andmaintain the growing-in-sizeaggregation network, whileoptimizing OPEX

•Capability of mixing optical andelectronic aggregationcapabilities, according to the realneeds, in order to optimize CAPEX

•QoS management for businessservices and for residential beyondInternet access

Metro

CPE (bus.)

BB access

CPE (bus.)

•Multi-service transportcapability, for converging IP/MPLStraffic coming from edge/outercore routers with clear channel(carrier’s carrier) type services,evolving to very fat pipes (10Gb/s⇒ 40Gb/s ⇒ 100Gb/s)

• OAM capabilities in line withtraditional transport services overgeographical distances

•Distributed restoration, in orderto fully exploit the meshed natureof core networks

•“Coarse” QoS, traffic andgranularity managementaccording to the nature ofaggregated streams in core network

Core

Aggregationtier 1

Aggregationtier 2

Edge


ADMADM ADMADMXC XC

Optical Access


Key transport networking needs: metro and access





Access

CPE (res.)





Metro

CPE (bus.)

BB access

CPE (bus.)





Core

Aggregationtier 1

Aggregationtier 2

Edge


ADMADM ADMADMXC XC

Optical Access

TMPLS allows for a convergent packet platform forany L2 protocol, thanks to a concept analogous toPWE3 as defined by IETF. On top of this, TMPLSadds OAM tools, linear and ring protection schemesand the possibility of a distributed control plane, all ofthem coming from transport experience, defined byITU and in common with other transport technologies.Multicast capabilities are as well under definition.

TMPLS allows for a convergent packet platform forany L2 protocol, thanks to the PWE3 concept definedby IETF. On top of this, TMPLS adds transportspecific OAM tools, linear and ring protectionschemes along with the possibility of GMPLSdistributed control plane, that is in common withother, TDM and λ based, transport technologies.Multicast capabilities are as well under definition.


Key transport networking needs: core





Access

CPE (res.)





Metro

CPE (bus.)

BB access

CPE (bus.)





Core

Aggregationtier 1

Aggregationtier 2

Edge


ADMADM ADMADMXC XC

Optical Access

TMPLS is an ideal packet platform foredge router interconnection,leveraging on MPLS pipes as natural“channels” of the router interfaces, tobe used in order to classify traffic andforward it through a TMPLS tunnel toits due destination. Besides commonwith other transport layers OAM andprotection like in metro, GMPLScontrol plane adds the possibility ofpath setup and restoration jointly atpacket/TDM/λ level. The conceptual,and then physical, distinction between“service” (IP/MPLS) and “transport”(TMPLS) allows for the design ofsystems the complexity (and cost) ofwhich is correctly tailored to transportneeds.


Network vision: aggregation networks layers usage

Packets, framesTransparency

TMPLS prot.Resiliency

TMPLSNetworking

Eth, ATM, FR ...Clients

TMPLS network (stat.mux.) layer

TMPLS for statistical multiplexing, SDH for transparent transport, λ for optical bypass

VCx, bit streamTransparency

SDH prot.Resiliency

VCx, VCx-ncNetworking

PDH, SDH, “Any”Clients

SDH network layer

STM-NTransparency

NAResiliency

NANetworking

STM-NClients

ODU adaptation (OAM) layer

OTU1, OTU2Transparency

Och prot.Resiliency

OchNetworking

OTU1, OTU2Clients

Och network layer

Och (λ)

XC

λ

λXC

XC XC

ODU

ODU

ODU

ODU ODU

ODU

SDH

TMPLS

XCReal networking

Pure adaptationλ

ODU


Network vision: core networks layers usage

TMPLS for statistical multiplexing, SDH and OTH (seen as ODU + λ) for transparent transport

Och (λ)

XC

λ

λXC

XC XC


TMPLS restor.Resiliency

TMPLSNetworking

PoS, Eth(Martini)Clients


VC4, framesTransparency

VC4 restor.Resiliency

VC4, VC4-ncNetworking

STM-N, EthernetClients

SDH network layer

STM-N, PHYTransparency

ODU restor.Resiliency

ODUNetworking


ODU network layer

(OTU1), OTU2Transparency

λ restor. (tba)Resiliency

λNetworking

(OTU1), OTU2Clients

Och network layer

ODUODU

ODU ODU

ODU

SDH

TMPLS

XCReal networking

Pure adaptationλ

ODU


Och (λ)

XC

XC

XCXC

XC XC

ODUODU

ODU

ODU

Core transport network: target vision

OTH progressively replacing SDH, integrated packet/ODU/λ networking


TMPLS/GMPLSRestoration

TMPLSNetworking

PoS, Eth(Martini)Clients


STM-N, PHYTransparency

λ+ODU /GMPLSRestoration

ODUNetworking


ODU network layer

(OTU1), OTU2Transparency

λ+ODU /GMPLSRestoration

λNetworking

(OTU1), OTU2Clients

Och network layer

TMPLS

XCReal networking

Pure adaptationλ

ODU


TMPLS within the target network vision

ADMADM

ADMADM

Layer 1 (TDM and/or λ)

XC

XC

XC

ADMADM

CEs

Network vision is based on multi-service node concept, able to support both residential(wireline/wireless) and business services, in which TMPLS technology is chosen as aconverged packet network able to support any service type (Ethernet, IP/MPLS, FR, ATM, TDMvia CES, …) along with GMPLS distributed control plane, common also for TDM and λ layers.

PE

Layer 3

TMPLS (layer 1.5)

L2(L3) businessservices

Residential services

CPE

Layer 2

Mobile services

CPE

ADMADM XC


6 Appendix: basics of GMPLS CP protocol


GMPLS versus MPLS: key points

UNI GMPLS defines

A UNI concept (thus easing overlay dynamic approach)

A I(nternal)-NNI concept

A E(xternal)-NNI concept (thus allowing for inter-working among different operators)

MPLS implies the concept of (internal) NNI only Note that hierarchical OSPF is not defined today neither for MPLS nor for GMPLS

GMPLS defines a link protocol LMP: it discovers which (user or network) interfaces are connected to a link and allows for

detecting which type of service (TDM, packet, λ) is available at a UNI

Control plane and data plane GMPLS allows for a separation of data plane and control plane

Only control interfaces are used to flood LSAs

In MPLS every data interface has to be also a control interface And thus flood LSAs

Path concept Unidirectional in MPLS, packet only Bi-directional in GMPLS, multi-layer


GMPLS versus MPLS: OSPF flavours

OSPF in general• OSPF is a link state routing protocols• OSPF advertises to all the node of the network its adjacencies via LSA (Link State Advertisements)• Every node builds up the entire network topology and evaluates the shortest path for each destination

OSPF

TopologydB

OSPF for IP• Data plane = control plane• LSA refers to all interfaces• One topology dB is built up

OSPF-TE

ControlTopology

dB

CSPF

OpaqueLSAdB

DataTopology

dB

OSPF for MPLS• Data plane ≠ control plane (in termsof routes, not of topology)

• Data plane can be constrained• OSPF builds the control topology dB• Opaque LSA are used to advertisedata plane ports but are notmanaged by OSPF

• CSPF builds up the data top. dB

OSPF-TE

ControlTopology

dB

CSPF

OpaqueLSAdB

DataTopology

dB

OSPF for GMPLS• Same as MPLS, with extensionsin order to:

• Describe non packet i/fs• Managing TE link bundlingfor scalability (N p2p i/fs areadvertised as 1)

• Transmitting several opaqueLSAs in 1 IP (OSPF) packet


GMPLS versus MPLS: RSVP-TE

RSVP-TE for MPLS

Defines unidirectional paths

The IP packet containing RSVP signaling hasthe path terminal node as DA (i.e. has tofollow the path)

In signaling N paths between A and B acompression can be done using one single IPmessage for all the relevant information(saving the header overhead)

RSVP-TE for GMPLS

Defines bidirectional paths (so the signalingtraffic and the number of state machines is½ of MPLS)

The IP packet containing RSVP signaling hasthe next path hop as DA (i.e. can follow adifferent route wrt a data packet to reachthe node)

In signaling N paths between A and B acompression can be done putting in a singleIP message only the path identifiers (andnot the paths characteristics that are alreadyknown)

A B

“Path” signalling

ACK ACK ACK ACK

ACK ACK ACK ACK

“ResV” signalling

RSVP-TE• Defines a finite state machine for each path that is set-up• The “Path” signal contains the route (ERO: Explicit Routing Object)• The “ResV” signal allocates the resources and distributes the labels• Path and ResV signals are output periodically to maintain the path (soft state)


GMPLS UNI concept

Servernetwork

Clientnetwork

Clientnetwork

GMPLS engine

MPLS engine

UNIUNI

RSVP-TEGMPLS

A B C D

The MPLS engine uses the [servertechnology] connection established by

GMPLS as forwarding adjacency

OSPF-TEGMPLS

NNI NNI

NNINNI

GMPLS connection (λ, TDM, packet)

The GMPLS engine asks a “loose source routed”[server technology] connection from A to D

through B and C. Reachability information betweenA and D has to be known “a priori”. Connectionbetween B and C is routed based on topologyinformation exchanged among GMPLS NNI’s.


www.alcatel-lucent.com

transport networks and transport mpls - intranet...

Documents