photonic network - apnoms

1

Photonic Network

Ken-ichi Sato [email protected]

Satoru Okamoto [email protected]

Tutorial

October 1, 2003APNOMS 2003, Fukuoka, Japan

NTT Network Innovation Laboratories

2

Part I : Photonic Network -Why It’s So Important?-Progress of Broadband Internet AccessP2P CommunicationBroadband ServicesAdvances in Transport Network TechnologiesFundamentals of Photonic NetworkPhotonic MPLS Router

Part II : Photonic Network Control and ManagementPhotonic Network Control and Management OverviewPhotonic Network Architecture OverviewIP over Photonic Network Architecture OverviewMPLS, GMPLS (MPLambdaS), and ASONGMPLS protocolsGMPLS managementInteroperability Test Events of Photonic Network Control

Outline

3

Progress of Broadband Internet Access

4

0

2,000

4,000

6,000

8,000

10,000

1 10

Seoul

Tokyo

Tokyo

Geneva

Dusseldorf

Paris

London

New York

New York Tokyo

Tokyo

ADSL Monthly Charge (Including ISP Charge)

Down Stream Speed (Mb/s)

Yen

per

Mon

th

1$ = 122 yen1U= 109 yen

SwisscomDeutsche TelekomFrance TelecomBritish TelecomAT&TVerizon CommunicationsCorea TelecomNTT GroupYahoo BB

Prices as of March-December 2002.

Data from Ministry of Public Management, Home Affairs, Posts and Telecommunications.

AsiaEurope

USA

5

0

2,000

4,000

6,000

8,000

10,000

1 10

Tokyo

Paris

London

New York

Cable Internet Monthly Charge (Including Cable Modem Lease Charge)

AsiaEurope

USA

Down Stream Speed (Mb/s)

Yen

per

Mon

th TelecolumbusFT CableTelewest CommunicationsAT&T BroadbandITS Communications

Dusseldorf

1$ = 122 yen1U= 109 yen



6

Leased Line Monthly Charge (34-45 Mbps)

Distance (km)

1$ = 122 yen1U= 109 yen

Deutsche Telekom (Digital 34MS)France Telecom (Transfix HD; 34 Mbps)British Telecom (Mega Stream34)AT&T (ACCUNET T45)NTT East (ATM Mega-Link)


0

1

2

3

10 100

Tokyo

ParisLondon

New YorkDusseldorf

AsiaEurope

USA

Mill

ion

Yen

per

Mon

th

Tokyo

Paris

London

New York

Dusseldorf


7

New Z

eala

nd

Nether

lands

0

1

2

3

4

Switz

erla

nd

Japan

Icel

and

USANorw

ayCan

ada

Denm

ark

Belgiu

m

UK

Hong Kong

Singap

oreSw

eden

Taiw

anAust

riaFin

land

Korea

Austra

lia

0

5

10

15

20

25

Japan

Korea

Belgiu

mHong K

ongTai

wan

New Z

eala

ndSin

gaporeUSA

Canad

aAust

ralia

Nether

lands

Norway UK

Icel

and

Swed

enAust

ria

Switz

erla

ndDen

mar

kFin

land

16

40

80

Price of Typical BB Connection As a % ofMonthly Household Income, April 2003

Price Per 100 kbps of Data per Month, US $,April 2003

T. Reynolds, ITU-T Promoting Broadband Workshop, Geneva, April, 2003

%$/

100

kb/s

Price of Broadband Connection

1(0.18)

1.6(0.29)

120(21.2)

8

Reasons for ADSL High-Penetration in Japan?

1. Low Subscriber Charges Stemming From:

● Competitive Environment- Local loop unbundling for ADSL and FTTH- Collocation

● Many ADSL Operators: about 50

2. Higher Speed Systems (up to 24 Mbps)

N.B. No Financial Support from Government

9

Why is FTTH so Cheap in Japan?

1. Low Installation Cost• Ratio of Apartment Inhabitants is High; about

40%.• High population density; 70% of the land is

covered with mountain• Aerial cable is used for the last mile (no digging)

2. Strong Competition• Competition between ADSL and FTTH• Competition among FTTH operators

3. Cost-effective Systems using Ether-based MediaConverter and B-PON.

10

Bandwidth Requirement

DSCQS Value (%)

108642 3

01020304050607080

・TV Quality: > 10 Mbps・ITU-R Broadcast TV Quality: > 8 Mbps・VHS Video Quality: > 4 Mbps

Quality and Necessary bandwidth

Broadcast TV Quality

VHS Video QualityWorse than VHSVideo, but acceptable

Worse than VHSVideo, tolerable limit

ITU-R Broadcast TV Spec.

Required Bandwidth for75 % of Video Samples Mbps

11

Available Speed vs. Transmission Loss for ADSLM

axim

um

Do

wn

stre

am S

pee

d 12

10

8

6

4

2

0706050403020100

Mb/s 12 M type

8M type

1.5M type

25 dB

Transmission Loss between Subscriber and SP building (dB)

12

Flets ADSL 8M Type From http://www.ntt-east.co.jp/flets/misc/adspeed.html

Measured Downstream Speed for 8 Mb/s ADSL

Mea

sure

d D

ow

nst

ream

Sp

eed

Mb/s

0 10 20 30 40 50 60 Transmission Loss between Subscriber and SP building (dB)

・： # of data is 1○： # of data is 2-4●： # of data is >5

Total # of data >4,200

13

1. More than 8-10 Mbps will be required to get TV qualitystreaming video services stably, and much higher isneeded to support SHD video quality services.

2. ADSL provides several mega bit per second only forlimited users who are very near (less than about onekilometer) the service provider’s office. If the price isthe same for all subscribers, unfairness in terms ofavailable bandwidth exists. This unfairness willbecome more and more tangible when bandwidthdemanding services proliferate.

Why FTTH ?

14

P2P Communication

15Data from White Paper 2002, Ministry of Public Management, Home Affairs, Posts and Telecommunications.

Trends in Internet Usage by Purpose

： 2000

： 2002

330%170%

+260%

230%

260%

Multiple Answers Permitted, Excerpt

16

120

100

80

60

40

20

0

Mb/s

Outgoing Traffic from the University of Wisconsin to the Internet

: Napster: Others

1999 May 1999 Sept. 2000 May 2000 Sept.

17

● About 3.4 % of Internet Users (about 99 mill.)● Application Used for File Exchange (multiple answers permitted)

　　 WinMX : 82.4 %

Winny : 22.8 %

Napster : 22.5 %

● Average Number of Downloaded Files per User is 162,

of Which 32.5% were Video.

● 45.1% of them uploaded; Published 124 Files on the

　Average

Increase in P2P Traffic

P2P File Exchange Statistics in Japan Data are at June 2003.

Copyright and Moral Issues Need to be Resolved before P2P canBecome More Widely Spread. Increase in Users Will Cause SignificantNetwork Traffic Load.

18

Lessons Learned form P2P

Abrupt Generation of Large Volume of Traffic

Increase of Large File-size Traffic

Different Traffic Patterns（Up-/Down Traffic )

- Instead of 10:1(down:up), just 3(or 2):1

Number or Users Increased with the AvailableSubscriber Access Speed

Usage Increased as HD Storage Capacity withAlways-on Connections Increased (Local Storeand Replay Enhances Usage)

Peer-to-Peer is More Than Sharing Music/Video Files;It Might Create the Way to a New Business Scheme.

19

Broadband Services

20

Traffic Increase:> x 2/12 monthsFirst Stage:

Introduction ofWavelengthrouting

OADM

Phot

onic

MPLS

OXC

Phot

onic

Burst

Electrical Processing

Traffic IncreaseStemming fromNew Services:x n/12 months

Moore’s Law: x2 /18 months

Second Stage:Introduction ofSuper DWDM(>1,000 l’s)

Year Broadband andUbiquitous NW

Expansion ofCommunications Services

Enhancement of Photonic NW Services and Performance

21

Real Time VLBI Measurement System

Radio Wave from Galaxy

RX

TX

Photonic2Networks

RXTX

TXTX

VLBI: Very Long Baseline Interferometory

Usuda

Nobeyama

KSP Miura

National Astronomical Observatory

Communications Research Labs.

KSP Tateyama

Kashima Space Communications Research Lab.

NTT MusashinoR&D Center

22

Medical Network

Uses: Medical Applications, Education and Amusement

A large hospital generates 80 Gb of data each day.

Medical Applications

PublicHealthCenter

Firehouse Drugstore

MedicalInformationDatabase

Hospital

In-home Care

MedicalRecordData

!Tele-Immersion > 10 Gb/s (Database, Simulation, Rendering)!SHD Motion Picture = 6 Gbit/sec!High-Vision Motion Picture = 1.2 Gbit/sec

Education and Amusement

EducationNetwork

ElectricalLibrary

Lecturer Student

Data base

ServiceProvider

Photonic Network

23

• FedEx is already a Terabit network– thousands of disks and tapes shipped daily– jitter and delay is pretty poor– cost for shipping tape approx. 0.000001¢/byte– current cost of sending data over fiber 0.001¢/byte

(1998 CANARIE Inc.)

Photonic Network

・Light speed・Low cost（nighttime）

Transport of package media

24

Photonic Backbone Network

・immediate・low cost

Digital Cinema

QUALCOMM’s Approach Expectation for Digital Cinema - 31,000 screens (USA) - 231 major titles (1997) - Cost for printing, $700 mill. ( $22,400/screen)

Digital Cinema will Spur - increase in smaller theaters - increase in number of copies - world-wide simultaneous release

25

Resolution comparisonHDTV-1080p/24 vs. 4K SHD Digital Cinema with 8M pixels

HDTV-1080p　1920 x 1080 pixels

4Kp-SHD　3840 x 2048 pixels

Standard TV720 x 480

26

Format comparison of Digital Cinema

Polaroid film

35mm film

3024

60mm filmLaser Printer

PrintingMedicine

2000

1000

0

35mmMovie

HDTV1080P720P Standard　

TV(480i)

HDTV(1080i)

TV

Still Images Motion Picture

0

3000

SHD Digital Cinema with8M（3840x2048） pixels

Temporalresolution(Frame/sec.)

Spatial resolution（lines）

Legacy media

Horizontal resolution is used for motion pictures.　4k motion picture means= 2000 scanning lines　2k motion picture means = 1000 scanning lines (@HDTV)

4K

2K

27

Digital Cinema Architecture Empowered by Photonic Technologies

Photonic Network275 ″ Screen3840x2048 pixels@24p, 96 Hz Refresh

Multi-format Decoder

Photonic MPLSRouter

Digital CinemaDistribution Server

300-400Mbps IPStream ofFile Transfer

35 mm FilmDigitizer

ArchiveCenter

Motion Jpeg(15:1)

G-Ether

G-Ether

28

Streaming Trials at Internet2TCP/IP based streaming experiments over large-scale, high-speed networks (1) Stable Streaming of SHD Digital Cinema (cope with long delay)

(2) Arbitration of Streaming Traffic by using MXQ (MaXimal Queuing) Mechanism

(3) Wide-Area Multipoint Live Streaming （Flexcast）: Connecting 3 sites using GEMnet and Abilene (Internet2)

Yokosuka Japan (NTT Lab.), Chicago, (UIC/EVL，NWU/StarLight), Los Angeles (USC)

SHD Digital Cinema Streaming : From Chicago to L.A (3,000km) 300Mb/s

GbE

OC12/ATM

Abilene

LightReef Z4

LightReef Z4

Cisco SL6509

Traffic Monitor

NTT Server

Application Traffic MonitorDecoderUSC/ Zemekis Center

StarLightNew Route 66

Route 66

Abilene　１０ GbE

<20 MbpsSunny Vale

PacificLink

NTT

GEMnet

Cisco 12404

29

Demonstration of SHD Digital Cinema in cooperationwith European Digital Cinema Forum (EDCF)

CineCitta in Rome–Famous Cinema Studio–Ministry of Culture is pushing

4K Digital Cinema in Italy

National Film Theater in London

–Testbed of EDCF is located atNational Film Theater (6/24)

–Support of DTI

30

Advances in TransportNetwork Technologies

31

• Less Service Dependent than Circuit/Flow• Grouped Circuits/Flows Serving as a Unit of Network Operation,

Design and Provisioning, Including Traffic Engineering• Object to Be Rearranged in Node and Transmission

Line/System Failure Restoration

Switch Switch

Cross-connect Path

Optical Fiber/Radio Wave

Optical Fiber/Radio Wave

Circuit

Cross-connect Path

Circuit

Role of Path

32

Switch

Higher-Order Electrical Path Cross-connect

Optical Fiber

Circuit

OpticalFiber

Higher-Order Electrical Path

Lower-Order Electrical Path Cross-Connect

Lower-orderElectrical Path

Hierarchical SDH Path Structure

ServiceAccess

Trans-Access

33

Service 1

(a) Media management by VPs

Transmission Link

Media 1

(c) QoS management by VPs

QoS 1QoS 2QoS 3

(b) Network service management by VPs

Transmission Link

Transmission Link

Media 2Media 3

Service 2Service 3

VP

VP

VP

8 7 6 5 4 3 2 1VPI

VPI VCIVCI

VCI PT CLPHEC

123456

53

OctetBit

Header

Information field

Header structure at NNI

Header Information field

5 octets 48 octets

53 octets

Cell structure

ATM Virtual Path

34

Virtual Path Benefits - Compared to Digital Path in STM -

Simplification of Interface and Node Structure

Simplification of Path Layer Architecture

Simplification of Path Accommodation DesignIn Terms of Path Hierarchy

Network Flexibility Enhancement

35

Hierarchical Path Structure

Switch/RouterFunction

Optical Path Cross-connect

Optical Fiber

Circuit/Flow

Optical Fiber

Optical Path

Electrical PathCross-ConnectFunction

Electrical Path(VP, LSP, DP)

36

Optical PathWP (Wavelength Path) and VWP(Virtual Wavelength Path)

WP 1 WP 2 WP 3 VWP 1 VWP 3VWP 2

VWP 4WP 4

37

1. Enhanced Transmission Capacity With WDM

2. Enhanced Cross-Connect Node Throughput with Wavelength Routing

3. Flexible and Progressive Transport Capacity Enhancement

4. Provision of Transport Platform(Different Degree of Transparency Can be Utilized)

5. Effective Network Protection/Restoration

Advantages of Optical Path

38

Multiplexing and Path Realization Technologies

MultiplexingTechnologies

Path Technologies

Path Identification(# of Paths/Link)

Soft/HardState Routing

PDHSDH

ATM

Packet

WDM

Digital Path(VC-1n, VC-3/4)

VP

LSP

OpticalPath

Time Position inthe TDM Frame

(< 192)

Cell header (VPI)(< 4096; NNI< 128; UNI)

Sim Label(<220)

Wavelength(<1,000)

Hard

Hard

Soft

Soft

Store-&-ForwardElectrical Processing

+Space Switch

Waveguide Router (Self-Routing)

and/orSpace Switch

Time SlotInterchange

+Space Switch

Store-&-ForwardElectrical Processing

+Space Switch

39

Label-Switching Router (LSR)

MPLS

Label Switched Path (LSP)

Egress LSRIngress LSRMPLS domain

IP Router

IP packet isencapsulated.

Label is swapped.

MPLS integrates IP and data-link layer technologies.

Removes theMPLS header.

(LabelEdgeRouter)

(Label EdgeRouter)

(LabelSwitchRouter)

40

The initial MPLS effort will be focused on IPv4. However, the coretechnology will be extendible to multiple network layer protocols(e.g., Ipv6, IPX, Appletalk, DECnet, CLNP). MPLS is not confinedto any specific link layer technology, it can work with any mediaover which Network Layer packets can be passed between networklayer entities.

MPLS provides connection-oriented (label based) switching basedon IP routing and control protocols. MPLS may be likened to a'shim-layer' which is used to provide connection services to IP andwhich itself makes use of link-layer services from L2 (e.g. PPP,ATM, Ethernet).

MPLS

A Framework for Multiprotocol Label Switching <draft-ietf-mpls-framework-05.txt> September 1999

41

Layer 3

Layer 2

IP Header

ATM

IP

VPI/VCI DLCI Shim Label

Frame Relay Ethernet

Shim Label

PPP

MPLS Labels; Label stacking and support of various media

MPLS is intended to run over multiple link layersSpecifications for the following link layers currently exist:

ATM: label contained in VCI/VPI field of ATM headerFrame Relay: label contained in DLCI field in FRheaderPPP/LAN: uses ‘shim’ header inserted between L2and L3 headers

42

MPLS Header

Layer 2 Header MPLS Header IP Packet

Label (20 bits) EXp (3) S (1) TTL (8 bits)

Label; 20 bits (4 bits for indicating how to handle labels, 16 bits for indicating FEC)Experimental (was CoS, class of service); 3 bitsStacking bit; 1 bit (indidcates the presence of a label stack; 1=last entry in label stack)Time to Live; 8 bits (same functionality as IP TTL; used to through away looping packets)

4 Bytes

★ short and fixed length★ associate to Forward Equivalent Class (FEC)

43

IP over ATM, and IP + ATM (MPLS)

Protocol

IP over ATM

Neighbor Routers All routers

Resource Management Per VC Reservation

IP + ATM (MPLS)

Routing protocol (e.g., OSPF)determines connectivity

Separate protocol for ATM(connection-oriented layer) and IP(datagram layer) routers （twolevel networks without integratedNM)

Existing Routing protocols (e.g.,OSPF) establish reachability(LSP),LDP establishes label to destinationnetwork mappings

MPLS “partition” allocated foreach link

Connection Full connection-orientedemulation

Hybrid Architecture(Connection-oriented (LSPs)plus datagram mechanism)

Hierarchical Forwarding VP + VC (IP+ATM)Arbitrarily nested (MPLS)VP + VC

44

Comparison of IP over ATM and MPLS (IP+ATM)IP over ATM MPLS

}Structure

Mapping ofIP Packetinto Layer 1

IP Router ATM-SWIP Router PartMPLS-SW Part

MPLS-Router

IP Packet

Segmentationinto ATM Cell

Accommodation into SDH Frame

IP PacketLabel AddedPPP/HDLC Processing

Accommodation into SDH Frame(Ethernet transmission is also possible w/oPPP/HDLC processing)

Characteristics

NetworkOperation

- VPI(/VCI) is swapped link by link- SAR Burden- Meshed VP connection between routers (Limited network expandability)

- Label swapping (link by link)- Variable length packet processing at MPLS-SW- Hierarchical LSPs with label stacking

- Separate (independent) IP and ATM layer operation (ATM will be common transport platform to other services than IP)

- Integrated IP and MPLS layer operation- Operation on different label switched path realization techniques- Standardized MPLS signaling protocols

45

☆Traffic Engineering (Path oriented)- Possibility to set-up other paths than “shortest paths”- Multiple paths between two points: Load sharing,

1 + 1, 1 : N, M : N Protection☆Very flexible because of de-coupling of forwarding　and routing (forwarding decision is separated from routing

process)☆Support capabilities of VPNs and new service provisioning

- by allowing the forwarding infrastructure to remain the same while new services are built through the assignment of packets to an LSP

☆Enahanced QoS

Why MPLS?● MPLS helps scaling pubic IP networks and enhance　network performances● MPLS provides network providers with means that can　differentiate their services from others.

46

2

Routing G-bit Networking Tera/Peta-bit Networking

IP

ATM

etc

.SD

H

Elec

trica

lO

ptic

al

Wav

e-le

ngth

Routers basedon softwarerouting

Routers basedon hardwarerouting(ASIC)

Tbit Routers(IP v6,

HierarchicalAddressing etc.)

Routing isdone withIP only(RouterMulti-hop)

Laye

r 2La

yer 1

PhotonicMPLS

Node ThroughputEnhancement

TrafficEngineering(QoSguarantee)

IP over Optical Path

IP over ATMIntroduction ofUnderlyingTransferMechanismwhich enableseffective trafficengineeringMesh-likeconnectionis possible(Routersingle hop)

MPLS: Multi Protocol Label Switching

MPLS

Router Throughput IncreaseIP over SDH IP over WDM (SDH)

Laye

r 3

Enhancement of Networking Function

Evolution of IP transport mechanism

47

MPLS and Photonic MPLSMPLS Router

Ingress Egress

Label Switch

Ingress Egress

Photonic Router

MPLS

Photonic MPLS

l Label is added to each packet.

l Wavelength label is added to each layer 1 stream.

WP approachVWP approachWavelength Label

Optical Label Switch

Labeled PacketLabeled Packet

Label

IP Packet IP Packet

IP PacketIP Packet

48

MPLS and Photonic MPLS Cont..

IP packet

MPLS Router

IP packet

Ingress

Labeled Packet

IP packet

Egress

IP packet

Ingress Egress

Photonic Router

MPLS

Photonic MPLSl Label merge function can be realized.

l Label merge is difficult. Label stack is also difficult.

IP packet

IP packet

Labeled Packet

49

MPLS and Photonic MPLSLSP 1

LSP 2

LSP 3

LSP 1

LSP 2 OLSP 1

MPLS-Router

Photonic MPLS-Router

OLSP 1

LSP1 and LSP 2 are accommodated within LSP 3.

LSP1 and LSP 2 are accommodated within OLSP1.

MPLS-Router MPLS-router pert multiplexes LSP1 and

LSP2 and connects to OLSP1.

50

Comparison of Electrical MPLS and Photonic MPLS

Electrical MPLS Photonic MPLS

Path Label Switched Path(Label is attached to each packet)

Optical Path(Label is attached to data stream)

Label Merge Yes Difficult

Label Stack Yes Difficult

Path State Soft Hard

# of Paths/LinkCan be very large

(220=1,048,576) Limited ( < 1,000)

Path Bandwidth Any Usually fixed and large (Gb/s)

Label Swapping Yes Yes (with wavelength conversion)

No (without wavelength conversion)

Hit-less RouteChange Yes (Make-before-break) No (possible only at electrical level)

51

PTSPhotonic (MPLS) Plane

IP (MPLS) Router

Photonic (MPLS) Router

MPLS Plane

AdministrationDomain

Photonic MPLS and MPLS hierarchical architecture

52

Best Effort(Connectionless) Engineered QoS

(Connectionless on Connection Oriented)

Guaranteed QoS(Connection Oriented)

ATM XC/Switch

MPLS RouterIP Router

Big IP RouterBig MPLS

Router

Photonic MPLS Router

Thro

ughp

ut (N

etwo

rk S

cala

bility

)


IP over ATM

OXC/OADM

Digital XC

53

OXCs and Photonic MPLS routers co-exist?

Yes, each has its own application.

Photonic MPLS routers are regarded as the ultimate inintegrated router systems.OXCs create optical platform on which different transfermode services can be provided.

Criteria of the Selection - Services (IP only or IP and other services, ex. Lambda service?) - Ownership (Are IP routers and OXCs and Transmission Equipment owned by the same provider?) - Segmentation of Network Management

54

This document describes extensions to MPLS signalingrequired to support Generalized MPLS. GeneralizedMPLS extends MPLS to encompass time-division (e.g.SONET ADMs), wavelength (optical lambdas) andspatial switching (e.g. incoming port or fiber to outgoingport or fiber). This document presents a functionaldescription of the extensions.

Generalized MPLS - Signaling Functional Description

Draft-ietf-mpls-generalized-signaling-00.txt

Genralized MPLS

GMPLS signaling: Extensions to MPLS signaling required to supportFour switching classes in GMPLS.

- GMPLS CR-LDP extensions- GMPLS RSVP-TE extensions

55

Enhancement of MPLS for GMPLSSupport different types of switches; packet-switchcapable, layer 2-switch capable, TDM capable,lambda-switch capable, fiber-switch capable.

A new Link Management Protocol (LMP) that runsbetween adjacent nodes and is used for both linkprovisioning and fault isolation.

Enhancement of OSPF/IS-IS routing protocols toadvertise availability of optical (and other)resources in the network.

Enhancements to RSVP/CR-LDP signaling protocolsto allow a label-switched path to explicitly specifiedacross the optical core.

Optical label switched path is bi-directional.

56

IP MPLS GMPLSMPLS-TE

Connection-lessConnectionOriented LSP(LSP, VP/VC,DLCI)

MPLS+TEExtensions

Multi-Layer LSPs+ TE Extensions

Connection

Routing/Forwarding

Hop-by-HopPacket Routing

Label Switchingwith LabelSwapping

MPLS+ExplicitlyRouted TE-Tunneling

Explicitly RoutedTE-Tunneling withMulti-layer LSPs

Control andData Plane Coupled

LogicallySeparated

Logically/PhysicallySeparated

Logically/PhysicallySeparated

Routing

Distributed IGPProtocol forRouting andPath Calculation

Hop-by-HopRouted Path:OSPF, ISIS,BGP-4

Constrained-based Routing:OSPF-TE, ISIS-TE, BGP-TE

Extensions toHandle Multi-LayerTE and Each LayerPath PeculiaritiesLink managementprotocol (LMP)

Signaling LDPRSVP-TE

CR-LDP,RSVP-TE

CR-LDP-EXT,RSVP-TE-EXT

Evolution of IP-oriented Network Technologies

IGP, RSVP

(Distribute Topology Information Only)

57

Fundamentals of Photonic Network

58

IX Backplane Maximum/Minimum Traffic Volume

The maximum and minimum daily traffic volume on the IX backplane in bits per second:

http://www.jpix.ad.jp/en/techncal/traffic.html

・IP backbone traffic is increasing by 160% every year (100 times in 5 years, 10,000 times in 10 years)

Internet Traffic Growth in Japan

x 2.6 /year

59

x 4.7 /year

The other IX (JPNAP) has higher traffic growth rate by370% every year .

60

Monthly aggregate traffic at AMS-IX, Nov 2001 - Jan 2003

By Erik Radius, SURFnetON*VECTOR photonics workshop, San Diego, February 3-4, 2003

x 2.6 /year

Internet Traffic Growth in Europe

61 61

・Traffic growth in NA was 85 % in 2002.

Internet Traffic Growth in North America

62

Characteristics of IP Traffic

Rapid and Unpredictable Traffic Growth- Data traffic grows very rapid- Data traffic demands are unpredictable

in terms of capacity and location.- Traditional capacity planning methods

for voice traffic are rendered obsolete.

Unpredictable Traffic Pattern Change

Distance Insensitive

Minimum Intervals in Service Provisioning- Need to be attained in response to the

unpredictable traffic demand.

Slow/Steady and PredictableTraffic Growth

Predictable Traffic Pattern Change

Most calls Terminate within LocalArea

Service Provisioning can be basedon Planning

IP Traffic Voice Traffic

63

Connection distance (km)

Frac

tion

of to

tal tr

affic Voice traffic

(744 k m average)

Internet traffic(2729 k m average)

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

0 to 500

1000 to 1

500

3000 to 3

500

7000 to 7

500

4000 to 4

500

6000 to 6

500

2000 to 2

500

5000 to 5

500

USA Connection Capacity Statistics

64From OFC2002 Short Course, Optical Internetworking Architecture, by Joseph E. Berthold, March, 2002

Traffic Model in North America

Total Generated Traffic = about 3 Tb/s- No Relation to Network Topology

Distance Bandwidth Product = about 10M Gb/sec-km- Obtained by Mapping Demands to Physical Topology- Parameter more Related to Capital Cost

Maximum Transmission Capacity = 1 Tb/s per Fiber (10Gb/s x 100l)

Maximum Node Throughput = 50-250 10G port

65

DCS 3/1 or Patch Panel

Patch Panel

Metro Inter-Office Ring; 622 M, 2.5 G




Some Hundreds of Rings Exist in a Large Metro Network

About 10 Nodes (About 10 Offices/Node)

Metro-Core Office

About 10 Nodes (About 100 Rings)

Tens of rings

Metro-Core Ring; 2.5 G

Long-Haul Ring; 2.5 G/10G

Partial-Mesh Connection for IP Core Network

WDM TransmissionEquipment is Omitted

Network Configuration Example in North America

66

Link Distance Distribution

21%

37%

13%

8% 8%6% 5%

3%

0%

5%

10%

15%

20%

25%

30%

35%

40%

0-20

0

200-

400

400-

600

600-

800

800-

1000

1000

-120

0

1200

-140

0

1400

-160

0

1600

-180

0

Link Length (km)

Coun

ts (%

)

Circuit Length Distribution

9%

20%

16%

12% 12%

15%

9%

3% 3%

0%

5%

10%

15%

20%

25%

0-10

00

1000

-200

0

2000

-300

0

3000

-400

0

4000

-500

0

5000

-600

0

6000

-700

0

7000

-800

0

8000

-900

0

Circuit Length (km)

Coun

ts (%

)

Link Distance and Circuit Length Distribution in North America (40 node Model)

From J-K Rhee et al., Proc. SPIE, ITCom 2002, vol. 4872

> 90%

67From A. Solheim, Business Briefing, World Markets Research Center

Link Distance and Circuit Length Distribution in Europe

68

Wavelength Routing on Photonic SuperhighwayWDM + (electrical) IP Router

Photonic MPLS

Existing Network

Photonic Network

WDM LinkIP Packet

IP Router

PTSWavelength Routing

IP Router

Wavelength Path

WDM Link

Traffic Jam at Node

69

Path and Round Trip Time Measurements

June 1998

http://www.caida.org/outreach/presentations/nanog9806/

Internet Hop Distance Distribution

70

End-to-End Node Cost Reduction

End-to-End node cost ratio“IP over photonic” to “IP over WDM”

PTS

WDMLT

Intermediate nodes

• IP over Photonic

• IP over WDM- 2.5 Gbit/s IP router I/F (OC-48c or STM-16)- 20 Gbit/s transmission (2.5 Gbit/s • 8 l)

Cost ratio per 2.5-Gbit/s capacity; PTS : IP router : WDM-LT

3.5 : 1.5 : 2.0

3.0 :0.75 : 2.0

1 2 3 4 5 6 7 8 9 10

1.0

0.8

0.6

0.4

0.2

Number of intermediate nodesIntermediate nodes

IP Router

IP Router

Edge node Edge node

Edge node Edge node

71

Comparison Between Ring and Mesh Network

Distance 1

Total Link Length for Ring Network Architecture = 32Total Link Length for Mesh Network Architecture = 10

Ring Network Transmission Cost／ Mesh NetworkTransmission Cost = 32/10 =3.2

Distance 1

72

Comparison Between Ring and Mesh Networks

2/4 Fiber Ring Architecture Mesh ArchitectureMinimum planning. Add capacityas needed (Pay as you growsolution). Hot-spot bandwidthupgrade.

Adaptability to dynamictraffic patterns (cannotplan 10 years anymore;IP traffic is unpredictable.)

Total throughput must be pre-planned and installed.

Bandwidth Scalability Limited (two fiber to four fiber up-grade and

multiple ring interconnection).

Controlled and managed growthis possible.

Network Scalability Limited (multiple ring arrangement).Controlled and managed growthis possible.

Restoration Speed - 50 ms < 1 s

Network Management Simple More complicated

Network ResourceUtilization

HighLower

Adaptability to distance-insensitive traffic pattern(internet traffic).

HighLow

73

1975 1980 1985 1990 1995 2000

WDM, OTDM

WDM

ETDM

ETDM

10,000

1,000

100

10

1

0.1

0.01

Tra

nsm

issi

on

Cap

acit

y ( G

bit

/s)

Doubling Every Year(traffic increase)

1,000

100

10

1

0.195

10,000

90 2000

CiscoAvici

Cisco

Cisco

Sys

tem

Th

rou

gh

pu

t (

Gb

/s)

Cisco,Pluris

Juniper

★ Server Bottleneck ★ Node Throughput Bottleneck

x2 /18months (Moore’s Law)

Traffic, Transmission Capacity, and Node Throughput

Traffic Increase Rate≒Transmission Capacity Increase> Router/Server Throughput Increase (Moore’s Law)

Year

Electrical Router

Doubling Every Year(traffic increase)

74

Photonic Network Technologies (OXC, Photonic MPLS Router)

Distributed Server Arrangement＋

High-speed Networking among Servers/Routers

＋Traffic Engineering（MPLS/GMPLS）

Cut-Through of Electric Routers（Introduction of Optical Path）

Role of Photonic Network Technologies

★ Server Bottleneck ★ Node Throughput Bottleneck

Traffic Increase Rate≒Transmission Capacity Increase> Router/Server Throughput Increase (Moore’s Law)

75

Path Speed(Gb/s/path)

Co

st/P

ort

OXC

EXC

(a) Path speed and nodecost

No

de

Th

rou

gh

pu

t

(b) Path speed and nodethroughput where switchhardware scale is fixed.

2.5 10 40

2.5Gx256 10Gx64 40Gx16

10Gx256

40Gx256

Path Speed(Gb/s/path)

OXC

EXC

Po

wer

co

nsu

mp

tio

np

er c

abin

et

EXC

OXC

Node Throughput

(c) Node throughput andpower consumption ofcabinet

Comparison of OXC and EXC

76

1,000

100

10

1

0.1

1995

10,000

1990 2000

Avici

Cisco

Cisco

Juniper

NTTNTT

CiscoFujitsu

Lucent

Juniper

Avici

JuniperHitachi

Procket

CiscoNEC

2005

Moor’s law(x 2/18 months)

Traffic Increase (x 2/12 months)

Year

System Throughput （Gb/s）

Progress of Router Throughput

Photonic MPLS Router (NTT)

Electrical Router

OXC

77

Williams Communications, ECOC 2002

Distribution of Number of Circuits for Backbone Carrier

DS1

DS 3

OC-3OC-12 OC-48

35 %

45 %

13 %

3 % 1 %

Number of Nodes: 16Number of Links: 29Node Degree: 3.6

78

Photonic Network Architecture

79

OADM : Optical Add/Drop MultiplexerOXC : Optical Cross-connect

OXC

l1

lN・・

OADM

l1 l2 l1 l3

l2l3

l1 l2 l1 l3

l2l3

~10

0Gb/

s1T~

10Tb

/s>1

0Tb/

s

Year

Evolution of Photonic NetworkTransport ofInformation Block withWavelength Label

Distributed Controlnetwork withPhotonic MPLS

CentralizedControl Mesh-type Networkwith OXC’S

OADM RingNetwork

Point-to-Point WDMTransmission

PhotonicMPLS Router

IP Routing

WavelengthRouting

IP Packets aremapped withinWavelengthLabeled BitStream

Photonic Router

Dyna

mic

Con

trol o

f Wav

elen

gth

Stat

ic W

avel

engt

hO

pera

tion

80

Photons and Electrons are Very Different

Both Photons and Electrons have the nature of a particleand of a wave, but they are very different.

Photons usually behave like a wave. - Photon has no mass. - Photon has no charge.Electrons usually behave like a particle. - Electron has mass. - Electron has charge.

A crucial difference is the degree of susceptibility inregard to interacting with others.There is no basic optical functional device equivalent tothe transistor.

81

Characteristics of Photons and Electrons

Is it possible to bend the direction of photon propagation in free space?

Yes, but huge mass is necessary: The mass of the sun can bend light by 1.74second; Einstein’s Theory of Relativity.

1.74’’

If dielectric constant is changed, direction of light propagation can be changed.

Dielectric constant can be changed and controlled by using, - mirror and mechanical means - electro-optic effect - thermo-optic effect, etc.

82

Characteristics of Photons and ElectronsIs it possible to confine/store photon?

Yes, but huge mass -40 million times that of the sun- is necessary to create aBlack Hole. But can not get photon back out of black hole.

No Optical Capacitor.

Is it possible to slow down photon propagation speed?

Yes, but at almost zero degree.

300,000 km/sec at normal temperature

~0 km/sec at almost 0 Kelvin

83

0.5 mm5 mm

10 mm

40 k cell/chip1 cell @ 10 Gb/s

8.5 m

v=6x10-4 mm3/cell

V=100 mm3/cell >105 v

Bare fiberStraight Line

Bare fiberCoiled 6 cm

125 mmf

V= 16000 mm3/cell >107 v

Semiconductor Memory(S-RAM)

Optical Fiber Delay Line

Volume of Memory

Si Memory Chip

84

10

100

1,000

10,000

100,000

1994

1995

1996

1997

1998

1999

2000 2001

10,000

100,000

1,000,000

155

620

2500Ce

ll Bu

ffer p

er 1

50 M

b/s

Path

Spe

ed

Cell

Buffe

r per

Lin

k

Year Link Speed (Mb/s)

A

B

(Not a technical limit)

Available Cell Buffer per Link in ATM Systems

85

IP over ATM

ATM XC WDM LT

MPLS (IP plus ATM)WDM LT

IP over Optical Path

WDM LTOXC

Photonic MPLS

WDM LT

IP over Digital Path

Scalability Global address IP signaling (distributed control)

Traffic engineering(Introduction of paths)Node throughputenhancement (Cut-through)

Large capacitytransmission

NecessaryAttributes

Photonic MPLSRouter

IP Router

MPLS Router

IP/MPLS Router

IP/MPLS Router

IntelligentDigital XC

IP Core Network Configuration

86

MPLSRouter

OXCOXC

SDH

SDH

GMPLS

SDHSDH

GMPLS

Distributed Control

Photonic MPLSRouter（Integration ofIP router andOXC function）

Optical PathPlatform（OXC/OADM:different transfermode servicesare supported）

ControlSystem （router-based approach）

Peer modelOverlay model (in case of centralized control;distributed control is also possible.)

Centralized Control(carrier oriented approach)

MPLSRouter

IP control plane

SDH network control plane

Optical transport networkcontrol plane(centralized/distributed)

IP-based control plane

IP router control plane

MPLSRouter

PhotonicMPLS2Router

PhotonicMPLS2Router

MPLSRouter

IP and photonic layer control

Different Service Network Architecture

87

Illustration of Global Network Architecture- Centralized and Distributed Functional Allocation Schemes -

Carrier Class NetworkPlatform

OXC, OADM

LimitedQoSNetwork

NetworkManagement(Policy basedControl)

NetworkControl(Signaling)

NetworkElement &NetworkElementOperation

DistributedCentralized

Distributed

Distributed

Centralized

Photonic MPLSIP Centric Network

88

Control Plane Drivers

– Reduce operational and capital expenses• More intelligence at the NEs can reduce manual intervention and

reduce operational cost• Open standard based products increase competition and drives cost

down• Convergence of different layers can cut both capital and operational

costs

– Move from voice to data dominated traffic• Traffic is more dynamic and difficult to predict and transport layer

needs to be more agile• As the transport network becomes more dynamic, and number of

elements and connections increase, distributed control will be moreeffective

89


90

1

2

M

1 2 N

M×1Optical Coupler

1×2 Optical Switch

MEMS(OFF)

MEMS(ON)

DC-SW

3D-MEMS SW2D-MEMS SW

3-stage Clos

N×2NSW

M×MSW

N×2NSW

M×MSW

N×2NSW

M×MSW

2N×NSW

2N×NSW

2N×NSW

Output

Different Switch Architecture

Inp

ut

Co

llim

ato

r

Inp

ut

Output Collimator

Substrate

Output Collimator

Input Collimator Mirror Array

Mirror Array

91

1 x 2(1) DC - SW N2 / L

2 x 2(2) 3 stage Clos

2 x 2(3) X-bar

N2 / L

N2 2 N2

2 N3/24 2 N3/28

l

O C

M×NDC-SW

(1)

1

l2

1lM

2

N

W-DMUX

W/C

W/C

W/C

W/C

W/C

W/C

W/C

W/C

W/C

●●●

●●●

●●●

●●●

M×NDC-SW

(2)

M×NDC-SW

(N)

O C

O C

1

2

N

W-TdW-Td

W-TdW-TdW-Td

W-Td

W-TdW-Td

W-Td

S-TdS-Td

S-TdS-TdS-Td

S-Td

S-TdS-Td

S-Td

1

2

N

1

2

N

M

NM x NMsinglespaceswitch

W-Td

W-Td

W-TdW-Td

W-Td

W-TdW-Td

W-Td

W-Td1

2

N ●●●

M×LSW(1)

N×NSW(1)

L×MSW(1)

(2) (2) (2)

(N ) (L) (N )

1

2

N

λ1

λM

S-Td

S-Td

S-TdS-Td

S-Td

S-Td

S-Td

S-TdS-Td

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

●●●

Necessary Hardware of Switch Architectures

Architecture Element SwitchNecessary Numberof Element Switch

Necessary Numberof Equivalent 1x 2Switches

(1) DC - SW (2) 3 stage Clos (3) X-bar

92

Switch Scale, N (N x N)

Nec

essa

ry N

um

ber

of

Un

it S

wit

ches

Necessary Hardware of Switch Architectures

1

10

100

1,000

10,000

100,000

1,000,000

10 100 1000

1ｘ2 S

W

1ｘN SW

1ｘ2 S

W

2ｘ2 SW

1ｘ2 S

W

1

Switch Scale, N (N x N)

Nec

essa

ry N

um

ber

of

Eq

uiv

alen

t 1x

2

Sw

itch

es

1

10

100

1,000

10,000

100,000

1,000,000

10 100 1000

DC-SW (8λ)DC-SW (32λ)3-stage Clos2D-MEMS (single stage)

3D-MEMS

DC-SW (8λ)DC-SW (32λ)3-stage Clos2D-MEMS (single stage)

3D-MEMS

93

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0.00 0.02 0.04 0.06 0.08 0.10

t/T（MTBF）

Prob

abili

ty

No. of failed mirrors24

16

8

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0.000 0.005 0.010 0.015 0.020 0.025

t/T（MTBF）

256×256 SW 1024×1024 SW

MTBF：MEMS mirror average life time

If the system life time is specified as 10 % failure of all mirrors, then about

50 % of systems show a life time of just 1.8 years when the MTBF of each mirror

including the feedback control system is 20 years.

Very reliable 3-D MEMS mirrors and control system are required.

Switch Reliability

Prob

abili

ty

0

No. of failed mirrors24

16

8

0

94

1X2 Optical Switch

1 2 n

12

m

OC OC OC

DC-SW: Delivery and Coupling SwitchW-DMX: Wavelength DemultiplexerOA: Optical AmplifierOC: Optical CouplerPLC-TOSW: Planar Lightwave Circuit -Thermo-optic Switch

OA OADC-SW

W-DMX

OCFiber Line 1

Fiber Line 2

Fiber Line N

DC-SW

DC-SW

DC-SW architecture

DC-SW　board

PLC-TOSW

Optical Cross-Connect Architecture

Link Modular Unit

Link Modular Unit

Link Modular Unit

OPXC Architecture

95

PLC-Switch

Silicon Substrate WaveguideCore

Cladding

Thin Film Heater

PLC: Planar Lightwave Circuit using Silica-based Waveguides

Ref.: T.Goh et al., Proceedings of SPIE, vol.4582, pp.49-56, APOC 2001, Beijing China.

Mach-Zehnder (MZ) 2x2 TO Switch

0 10 20 30 40Time (msec)

Applied Power

Optical Output

ON OFF ON

0.35 W-0.5 W

2 ms

Switching Response

96

Photonic MPLS Router Architecture

OXC

PAD

Photonic Router Hardware

Single ChannelLinks (UNI)

NNI : Network Node InterfaceUNI : User Network InterfacePAD : Payload Assembler/Disassembler

(UNI from/to NNI converter)NE-Mgr : Network Element Manager

(Photonic router software)DCC : Data Communication ChannelOSC : Optical Supervisory Channel

OpticalRepeater

Uni-directionalWDM Links

(NNI)

NE-Mgr

DCC over OSC

IP packetIP packet

Ingress

C ontrolmessage

97

MPLSRouter

LRU

Optical amplifier

WavelengthDemultiplexer

Optical coupler

DWDM transmission fiber

Opticalswitch

IP router/Ethernet switchConnection fiber

Wav

elen

gth

con

vert

er

IP Controller MPLS Controller

MPlS Controller Photonic router network manager

PhotonicMPLSrouter

Manager

Opticalamplifier

DWDM transmissionfiber

IP router/Ethernetswitch Connection fiber

Co-operationbetween OXCand MPLSrouter.

Optical switchesfabricated byPLCtechnologies.

Dynamic opticalpath setup/tear-down.

Configuration of Photonic MPLS Router

98

Outlook and Specifications of Photonic MPLS Router

Item

Throughput

Systemthroughput

UNI

Optical switcharchitecture

Optical switch

Wavelength band

Optical channelspeedNumber ofwavelengths

Number of fiberports

Total switch scale

Scalability

MPLS routerscalability

Specifications

More than 5Gpps (Obtained withwavelength routing and MPLS router)

Maximum 2.56 Tbit/s

POS, ATM, GEther, etc.

Delivery and coupling type

Planar Lightwave Circuit (PLC)thermo-optical switch

1550 nm band (C-band)

2.5 Gbit/s (up gradable to 10 Gbit/s)

32 per fiber

8 input /output pairs (fiber port canbe added one by one)

256 x 256 channels

T h e n um ber of av a ilable o p t i c a lchannels is expandable up to 256, with 8wavelengths’ modularity (each switchmodule accommodates 8 wavelengths.)

Maximum number of available POSinterface is 128. Consists of one totwenty MPLS routers.

Part II : Photonic Network Control and Management

• Photonic Network Control and Management Overview

• Photonic Network Architecture Overview• IP over Photonic Network Architecture

Overview• MPLS, GMPLS (MPLambdaS), and ASON• GMPLS protocols• GMPLS management• Interoperability Test Events of Photonic

Network Control

Page 100

Outline

• Photonic Network Control and Management Overview– Target Applications– Standardization Partner-ship

• Photonic Network Architecture Overview• IP over Photonic Network Architecture Overview• MPLS, GMPLS (MPLambdaS), and ASON• GMPLS protocols• GMPLS management• Interoperability Test Events of Photonic Network

Control

Target Photonic Network Applications

• Traditional Photonic Network(for Telephone Network Backbone)

– Provisioning Base• Centralized NMS base

– Centralized Control Plane (C-Plane)– Centralized Management (M-Plane)

• New Photonic Network(for Data Network Backbone)

– Dynamic/Adaptive Network– Distributed C-Plane– Centralized M-Plane

Page 102

Network Reference Model

• Transport Network is constructed with 3 planes.– Data (or Transport) Plane– Control Plane– Management Plane

Control Plane

Data PlanePE

PE

PE

PE

PE

PE

UNIUNI

Management Plane

CICI

CMICMI MIMI

CECE

CE: CustomerCE: Customer’’s Equipment, PE: service Providers Equipment, PE: service Provider’’s Equipments Equipment

Service Provider A Admin Domain

UNI E-NNI (b)

Service Provider B Admin Domain

Domain A1 Domain A2

E-NNI (a)

I-NNI

I-NNII-NNI

Provider A has divided network into multiple control domains

(e.g., vendor, geographic, technology, business unit, etc.)

Provider B’s network is a single

control domain

UserDomain

User-Network Interface (UNI):operations between user and service

provider control domains

Exterior Network-to-Network Interface (E-NNI):

inter-control domain operation.(a) multi-domain operation for a

single service provider(b) multi-domain operation among

different service providers

Interior Network-to-Network Interface (I-NNI):

intra-control domain operation

Network-to-Management Interface (NMI):

operations between management systems and service provider administrative domains

Fundamentals - Reference Points

firewall

firewall

L2/L3

L2/L3

LoadBalancer

LoadBalancer

Page 104

Traditional (Telephone Network)

• Centralized network operation and management with Network Management System (NMS) located in the network operation center (NOC).

– Labor-intensive processes• Error-prone, slow, and high operations costs• Rapid provisioning or network reconfiguration for resource optimization is

difficult (if needed).– There is no clear distinction between C-Plane and M-Plane.

Every action flows through the central NMS.• Limited scalability, visibility and manageability

Data PlanePE

PE

PE

PE

PE

PE

UNIUNI

CI & MI

NOC

CE CE

NMS C+M C+M -- PlanePlane

New (for Data Network)

Control Plane (distributed)

Data PlanePE

PE

PE

PE

PE

PE

UNIUNI

CICI

CMICMI

MIMI

• Intelligent Photonic Network– M-Plane (NMS) and C-Plane are separated.– Neighbor Discovery– Routing/Topology Dissemination– Connection Signaling

• Automated Processes, Scalability, Robustness, Efficiency

NMS (monitoring)MM--PlanePlane

Page 106

Intelligent Photonic Network Foundations

• Protocol Functions– Discovery

• Neighbor and link identity and characteristics

– Routing/Topology Dissemination• Network topology and resource availability

– Connection Signaling• Automated provisioning and failure recovery

• Concepts endorsed by every standards body– ITU-T, IETF and OIF

Standardization Bodies and Forums

ITUITU International Telecommunication UnionInternational Telecommunication UnionIETFIETF Internet Engineering Task ForceInternet Engineering Task ForceOIFOIF Optical Internetworking ForumOptical Internetworking ForumTMFTMF TeleManagementTeleManagement ForumForum

ASONASON MPLS, MPLS, GMPLSGMPLS

Optical UNI/NNIOptical UNI/NNI

Focused in this tutorial

NMSNMS

Page 108

• Different focus– ITU focuses on architecture

• ASON architecture– IETF focuses on building blocks

• GMPLS protocol specs.– OIF focuses on applications and interoperability

• Implementation Agreements (IAs) and Interoperability Test Events

• Common goal: better photonic networking • Recognized need for coordination

ITU vs. IETF vs. OIF

L. Ong, “Optical Control Plane Activities in IETF and OIF”, ITU-T Workshop on IP/Optical Chitose, Japan, 9-11 July 2002

Major Work Areas of ITU (ITU-T SG15)

• Optical Transport Network (OTN) structure – Automatic Switched Optical Network (ASON)– Architecture and interfaces for the OTN– Optical (Photonic) Cross-Connect and Switch functions– Network management and control

• OTN technology (terrestrial and submarine)– Coarse and dense WDM, STM-256 (40Gb/s) signal

channels– Optical components & amplifiers (e.g. tunable filters)– Fiber characteristics, more channels/fiber– Transmission technology (Soliton/RZ), long reach

Page 110

OTN Technology Standardization Work Plan

• Standardization areas covered– OTN Technologies (variety of aspects)– SDH & SONET

• Next Generation data centric transport network– Generic Framing Procedure (GFP)– Virtual Concatenation (VCAT)– Link Capacity Adjustment Scheme (LCAS)

– OTN Transport Plane (D-Plane)– ASON Control Plane (C-Plane)

• IETF’s Traditional Focus– The Internet: IP and IP Services – routing,

transport, applications, security & management• Sub-IP Area

– Coordinates activities below the IP layer, esp. MPLS/GMPLS and IP over Optical.

IETF Optical Standards

IESGIESG

Transport RoutingApp., General, Internet, OAM,

SecuritySub-IP

sigtran pwe3 ospf idr ipo ccamp

ipo : IP over Opticalccamp : Common Control and

Measurement Plane

IESG : The Internet Engineering Steering Group

Areas

WGs

Page 112

IETF GMPLS: History

• How did GMPLS start?– Outgrowth of MPLS - IP traffic engineering work – “Generalized” protocols for label-switched path

creation• Fiber switching• Wavelength/Waveband switching• Time slot switching (SDH/SONET)

• Possible Architectures – Flat network – routers and optical systems fully

peered (Peer Model)– Hierarchical network – routers are optical clients

(Overlay Model)• Scope

– Support of IP networking over optical transport– Non-IP-related use of GMPLS is out-of-scope

Optical Internetworking Forum (OIF)

• The only industry group bringing together professionals from the data and optical worlds.

• Open forum– international– carriers– component and systems vendors– testing and software companies

• Launched in April of 1998 with an objective to foster development of low-cost and scaleable internet using optical technologies.

• Mission: To foster the development and deployment of interoperable products and services for data switching and routing using optical networking technologies.

Background and MissionBackground and Mission

Page 114

OIF Focus

• Low-cost Scaleable Optical Internetworking– IP-Over-Switched Optical Network Architecture– Physical layer

• Low-cost optical interfaces between networking elements• Standard device level electrical interfaces for low-cost

systems– Control layer interoperability between data and

optical layers• Dynamic configuration using IP signaling and control

mechanisms• Accommodate legacy network under the new

physical and control layer mechanisms

OIF Current Work

• UNI 1.0 (now developing 1.0R2)– Focus on SDH/SONET environment– Approved, interoperability events sponsored at

SuperComm2001 and OFC2003

• UNI 2.0– Extensions for, e.g., Ethernet environment, multi-homed

access, non-disruptive connection modification, enhanced security

• NNI– Interface between domains; but intra-carrier E-NNI (a)– InterDomain signaling– InterDomain routing (new hierarchical routing)

Page 116

Outline


• Photonic Network Architecture Overview– New Optical Layer Hierarchy (OTN) and SDH/SONET (Pre-

OTN)• IP over Photonic Network Architecture Overview

– Peer Model, Overlay Model• MPLS, GMPLS (MPLambdaS), and ASON• GMPLS protocols• GMPLS management• Interoperability Test Events of Photonic Network

Control

SDH and SONET

• Good for aggregating small flows into a fat pipe (First developed for telephone networks)

• Electric endpoints, strong protection, troubleshooting functionality

SDH: Synchronous Digital Hierarchy (ITU)SDH: Synchronous Digital Hierarchy (ITU)SONET: Synchronous Optical SONET: Synchronous Optical NETworkNETwork (ANSI)(ANSI)

SDH/SONETSTM-1 156 Mb/sSTM-4 622 Mb/sSTM-16 2.5 Gb/sSTM-64 10 Gb/sSTM-256 40 Gb/s

VC-350 Mb/s

VC-11

1.5 Mb/s

VC-4VC-4-4c

600 Mb/s

150 Mb/s

VC Virtual ContainerSTM Synchronous Transport Module

Page 118

SDH Standardization

Network Architecture(G.803, G.805)

Structures and Mappings(G.707)

Physical Layer(G.957, G.691)

Equipment Functional Spec.(G.783, G.806)

Equipment Management(G.784, G.7710)

Information Model(G.774 Series)

Protection Switching(G.gps, G.841, G.842)

Laser Safety(G.664)

Data and Signaling Communications Network

(G.7712)

Jitter and Wander Perf.(G.825)

Error Performance(G.826-829)

Ghani Abbas and Stephen Trowbridge, “OTN Equipment and Deployment in Today’s Transport Networks”, ITU-T Workshop on IP/Optical Chitose, Japan, 9-11 July 2002

Optical Transport Network (OTN)

• Metro Access• SDH metro ring applications• Multi-Service Provisioning Nodes - combining data and SDH

• Metro Core• SDH ADM metro ring and mesh application• Optical add/drop multiplexers (proprietary)

• Long Haul/Ultra Long haul• SDH ADM rings and line systems• DWDM line systems (proprietary)

TodayToday’’s Transport Network Environments Transport Network Environment

New Generation Transport Network is required New Generation Transport Network is required OTN

Page 120

OTN Requirements

• Functionality as that offered by SDH or better• Transparent transport of SDH and other payloads• Stronger FEC (forward error correction)

ITU-T G.709 is the answerG.709 defines the interfaces for the OTN

New Transport Network Technologies are called OTN.New Transport Network Technologies are called OTN.So, old fashioned SDH/SONET are called PreSo, old fashioned SDH/SONET are called Pre--OTN.OTN.

Optical Transport Network (OTN) Standardization

Network Architecture(G.872)

Structures and Mappings(G.709)

Physical Layer(G.692, G.959.1, G.694.x)

Equipment Functional Spec.(G.798, G.806)

Equipment Management(G.874, G.7710)

Information Model(G.874.1, G.875)

Protection Switching(G.gps, G.otnprot)

Laser Safety(G.664)

Data and Signaling Communications Network

(G.7712)

Jitter and Wander Perf.(G.8251)

Error Performance(G.optperf)

Ghani Abbas and Stephen Trowbridge, “OTN Equipment and Deployment in Today’s Transport Networks”, ITU-T Workshop on IP/Optical Chitose, Japan, 9-11 July 2002

Page 122

Optical Layer NetworkOptical Channel LayerOptical Multiplex Section LayerOptical Transmission Section Layer

Physical Medium

Digital Clients

Optical LayerNetwork

Fibre

NBT(The Next Big Thing!)

Multi-layer Transport Networks

IP

ATM

SDH

Peter Huckett, “Relating Optical Layer and IP Client Performance”, ITU-T Workshop on IP/Optical Chitose, Japan, 9-11 July 2002

IP Over OTN Structure

ATMHDLCEthernet MACRPR MAC

10GbELAN PHY

10GbEWAN PHY

GFP

SONET / SDH

G.709 OCh (optical channel)

Optical fibre / G.652, G.653 etc.

IEEE 802.2 LLC PPP AAL5

GbEPHY

IEEE 802.2 LLC

RPR PHY

IP

RPR Residential Protection RingHDLC High level Data Link Control procedure

Tobey Trygar, “Network Performance (IP/Optical) IP/Optical Performance Management”, ITU-T Workshop on IP/Optical Chitose, Japan, 9-11 July 2002

Page 124

IP over Photonic Network ArchitecturePeer Model and Overlay Model

• Peer Model – Flat network

• IP routers and photonic network systems fully peered

• from the view point of C-plane, all network equipments are categorized with IP routers.

• Overlay Model– Hierarchical network

• routers are clients of the photonic network• Clients of the photonic network are not limited

to IP routers: e.g. SDH box, ATM box, …

Peer Model

Photonic Network IP NetworkIP Network

Same Control Protocol: i.e. GMPLSSame Control Protocol: i.e. GMPLS

GMPLS GMPLSGMPLSGMPLSGMPLSGMPLSGMPLSGMPLSC-Plane

• Only one C-plane• In the C-plane network, IP network element

controllers and Photonic network controllers do not have a client-server relationship.

Page 126

Overlay Model

Photonic Network IP NetworkIP Network

IP network Control ProtocolIP network Control Protocol

GMPLS GMPLSGMPLSGMPLSC-Plane

• Multi C-Planes.• Photonic network control protocol can alter from IP

network control protocol.– It is possible to adapt centralized management system.

Photonic network Control Protocol

GMPLSGMPLSGMPLSGMPLSC-Plane Inter-layer signaling protocole.g. OIF UNI signaling

Outline


• Photonic Network Architecture Overview• IP over Photonic Network Architecture Overview• MPLS, GMPLS (MPLambdaS), and ASON• GMPLS protocols

– Signaling– Routing– Link management

• GMPLS management– MIBs

• Interoperability Test Events of Photonic Network Control

Page 128

Overlay vs. Peer

• IETF focused on the peer model– IP network should be constructed with Routers

and Links.– All routers (and also links) should be controlled by

same protocols (routing protocols, signaling protocols, and management protocols).

– Links are provided by transport networks.• Transport networks should be IP router empowered

• ITU and OIF focused on the overlay model– Photonic network should be constructed for

commonly used service platform.

What is the new services?What is the new services?

ASON (Automatically Switched Optical Networks)ASON (Automatically Switched Optical Networks)

GMPLS protocols are commonly used in IETF, ITU-T and OIF.

Press Release (4 December 2001)

The 40Gigabit per Second Phone Call:

Global Standards for Automatically Switched Optical Networks Enable New Market Services

Page 130

Drive to Automatically Switched Network

• Make the network intelligent• On-demand bandwidth to the edge of the

network• New applications

– Disaster Recovery – Distributed SAN– Data warehousing

• Backup Bunkers (no more tapes)– Big Pipes on Demand

• Download movies to movie theaters• Site replication

– Optical VPN – Grid Computing

• Different focus– ITU focuses on architecture

• ASON architecture– IETF focuses on building blocks

• GMPLS protocol specs to ASON.– OIF focuses on applications and

interoperability• Implementation Agreements (IAs) and

Interoperability Test Events to ASON.

Reminder (ITU vs. IETF vs. OIF)

Page 132

IP, MPLS and GMPLS

1. IP: Shortest Path takes all packets

2. MPLS: Traffic Engineering allows flows to be mapped to different paths for better utilization

3. GMPLS: MPLS control protocols could also set up connections in a circuit network

L. Ong, “Optical Control Plane Activities in IETF and OIF”, ITU-T Workshop on IP/Optical Chitose, Japan, 9-11 July 2002

MPLambdaS (MPλS) and GMPLS

• Multi-Protocol Lambda Switching: October 1999– Combining MPLS Traffic Engineering Control With Optical

Crossconnects (OXCs)1. provide a framework for real-time provisioning of optical channels in

automatically switched optical networks2. foster the expedited development and deployment of a new class of

versatile OXCs3. allow the use of uniform semantics for network management and

operations control in hybrid networks consisting of OXCs and labelswitching routers (LSRs).

– The proposed approach is particularly advantageous for OXCsintended for data-centric optical internetworking systems.

• This concept can be extent to other circuit switch systems.– Attempt to SDH control: March 2000– Generalized Multi-protocol Label Switching (GMPLS): June 2000

Page 134

GMPLS controllable equipments

• Packet-Switch Capable (PSC) : MPLS Router• Time-Division Multiplex Capable (TDM) : SDH (VC)-

XC• Lambda-Switch Capable (LSC) : OXC, PXC

– Include both single wavelength switch and multi-wavelength (waveband) switch

• Fiber-Switch Capable (FSC) : PXC

• Layer2-Switch Capable (L2SC) : ATM-SW, FR-SW, etc. [GbE-SW (?)]– Defined. But not implemented.

GMPLS LSP Hierarchy

LSP（Fiber Switch Basis）

LSP（Wavelength Switch Basis）

LSP（TDM Switch Basis）

LSP（Cell/Packet Switch Basis）

Fiber

Wavelength (Group)

Time slot

Cell/Packet

Page 136

GMPLS protocols

GMPLS signaling Protocol:Extensions to MPLS signaling required

to support five switching classes (PSC, TDM, LSC, FSC, and L2SC) in GMPLS.

GMPLS CR-LDP extensionsGMPLS RSVP-TE extensions

Link Management Protocol (LMP):runs between neighboring nodes is used to manage traffic engineering

(TE) links.will be used to maintain control

channel connectivity, verify the physical connectivity, correlate the link property information, and manage link failures.

IP based Control Plane

GMPLS routing protocol:D-Plane topology discoveryPath route selectionSwitching capability advertisement

OSPF-TE GMPLS extensions

Focused in this tutorial

Hierarchical LSP and Signaling (1/2)

Ayan Banerjee, et al., “Generalized Multiprotocol Label Switching: An Overview of Signaling Enhancements andRecovery Techniques”IEEE Communication Magazine, Vol. 39, No. 7, July 2001.

Packet (PSC) LSP1

λ (LSC) LSP3

Fiber (FSC) LSP4

Time slot (TDM) LSP2

LSC

FSC

Page 138

Hierarchical LSP and Signaling (2/2)

Ayan Banerjee, et al., “Generalized Multiprotocol Label Switching: An Overview of Signaling Enhancements andRecovery Techniques”IEEE Communication Magazine, Vol. 39, No. 7, July 2001.

PSC LSP

TDM LSP

LSC LSP

FSC LSP

GMPLS RSVP-TE signaling

OIF OUNI and IETF GMPLS Inter-working

OUNI Resv

ResvConf

ResvConf+ MESSAGE_ID_ACK

ACK

UNI Transport Connection EstablishedSource UNI-C may start transmitting

Destination UNI-C may start transmitting

ACK

R1 XC2XC1

OUNI Path

ACK S=R1, D=R2

S=R1, D=R2

I-NNI Path

OUNI Resv+ MESSAGE_ID_ACK

OUNI Path

ACK

S=R1, D=R2

I-NNI ResvMESSAGE_ID

MESSAGE_IDI-NNI ResvConf

R2OUNI OUNIGMPLS

Page 140

UNI - E-NNI Inter-working

Path

Path

Rev

Resv+ MESSAGE_ID_ACK

ACK

NNI Transport Connection EstablishedSource NNI may start transmitting

ACK

Source UNI-C Destination UNI-CUNI-N I-NNI

ResvConf

ResvConf+ MESSAGE_ID_ACK

ACK

Destination NNI may start transmitting

PathACK

ACKResv

ACKResvConf

I-NNI E-NNI E-NNI I-NNI I-NNI UNI-N

Vendor Vendor AA Vendor Vendor BB

GMPLS Management (MIBs)

• Link Management Protocol Management Information Base– draft-ietf-ccamp-lmp-mib-06.txt

• lmpNbrTable, lmpControlChannelTable, lmpControlChannelPerfTable, lmpTeLinkTable, lmpLinkVerificationTable, lmpTeLinkPerfTable, lmpDataLinkTable, lmpDataLinkPerfTable

• Generalized Multiprotocol Label Switching (GMPLS) Label Switch Router Management Information Base– draft-ietf-ccamp-gmpls-lsr-mib-00.txt

• gmplsInterfaceConfTable, gmplsInSegmentTable, gmplsOutSegmentTable, gmplsLabelTable

• Generalized Multiprotocol Label Switching (GMPLS) Traffic Engineering Management Information Base– draft-ietf-ccamp-gmpls-te-mib-00.txt

• gmplsTunnelTable, gmplsTunnelHopTable, gmplsTunnelARHopTable, gmplsTunnelCHopTable, gmplsTunnelPerfTable, gmplsTunnelErrorTable

Page 142

Outline


• Photonic Network Architecture Overview• IP over Photonic Network Architecture Overview• MPLS, GMPLS (MPLambdaS), and ASON• GMPLS protocols• GMPLS management• Interoperability Test Events of Photonic Network

Control– OIF (OUNI, E-NNI)– MPLS Forum (GMPLS signaling)

SuperComm2001 OIF UNI demo

• 25 vendors– 14 client equipment (e.g. IP router)– 14 optical equipment (e.g. OXC)

• Some vendors are software only testing.

Vendor XVendor X Vendor YVendor Y

Vendor ZVendor Z

CC--Plane NetworkPlane Network

Page 144

Interoperability Results

A B C D E*2 F G H I J K L*2 M N1 H H *1

2 H H3 *1 *1 *1

4 H H *1

56 *1

7 H H H8 *1

9 *1 *1 *1

10*2 H H11 *1

12 H H H H H H *1 H H *1 H13 H14 H *1

*1 - Not enough time to complete testing*2 - Proxy or test device supports control plane only.H = Another client was used to form Hetrogeneous testResults scrambled to protect individual companiesTested, Signaling & Transport success

ONE Device

Clie

nt D

evic

e

Color KeyNot Interoperable

Tested, Signaling success

•Over 196 Tests Conducted

99.5% Signaling Inter-working

68.6% Signaling & Transport Inter-working

OFC2003 UNI/E-NNI demo

• 12 vendors, 15 systems.• UNI GMPLS E-NNI GMPLS UNI

Page 146

GMPLS Interoperability demonstration

• Held at UNH Interoperability Lab• Staging for GMPLS demo at NGN 2002• Organized by The MPLS Forum• Participants

– Equipment ImplementationsRouters: Cisco, JuniperSwitch: Sycamore

– Emulated ImplementationsStacks: Netplane, DCLTest Eqpt: Agilent, NetTest

Overview of GMPLS Interop

Page 148

Thank you !

photonic network - apnoms

Documents