Beyond 10 GbE – Looking Ahead Qwest Communications International Mark Stine, CTO Government Services Division February 2005

Beyond 10 GbE – Looking AheadBeyond 10 GbE – Looking Ahead

Qwest Communications InternationalQwest Communications International

Mark Stine, CTO

Government Services Division

February 2005February 2005

Qwest Proprietary and ConfidentialQwest Proprietary and Confidential © 2004 Qwest Communications International Inc. © 2004 Qwest Communications International Inc.

2Some of the Panel’s QuestionsSome of the Panel’s Questions

What problem are we trying to solve ?

How does price/technology/demand determine next step beyond 10 Gbps ?

Where are we now ?

40 Gbps/ OC-768 or 100 Gbps ?


3What problem are we trying to solve? For Qwest What problem are we trying to solve? For Qwest any new technology must result in:any new technology must result in:

Revenue improvement• Enable new servicesEnable new services• Product line consolidationProduct line consolidation• Increase market coverage; provide services fasterIncrease market coverage; provide services faster• Manage disruption of legacy revenueManage disruption of legacy revenue

Cost reduction• Increase utilizationIncrease utilization• Reduce ports; minimize metallic interfacesReduce ports; minimize metallic interfaces• Increase power efficiency and equipment densityIncrease power efficiency and equipment density• Minimize revisits to equipmentMinimize revisits to equipment• Remote operationRemote operation

Operations simplification• Convergence, integration of networksConvergence, integration of networks• Ease provisioning, network configuration changesEase provisioning, network configuration changes• Improve inventory managementImprove inventory management• Reduce engineering and planning complexityReduce engineering and planning complexity

Business goals driven by – demand, capital, competition, relationships


4Qwest Multi-service, Packet Centric InfrastructureQwest Multi-service, Packet Centric Infrastructure

Other Carriers

Last Residential Mile

Last Business

Mile

MPLS/IP Core

Multi-service Delivery Edge

, sub- Edge

Optical Connectivity

OC-n, Ethernet Edge

SCPSCP

SBCSBC

ASASService Data

& Control

DW

DM

DW

DM

DW

DM

DW

DM

DW

DM

DW

DM

NG

-AD

MN

G-A

DM

NG

-AD

MN

G-A

DM

Broadband AccessBroadband Access Pipes• Optical, Packet, Wireless

Broadband AccessBroadband Access Pipes• Optical, Packet, Wireless

Physical Domain(next gen optical)

Physical Domain(next gen optical)

Next Generation OSSOSSNext Generation OSSOSS

Control/Applications DomainControl/Applications Domain(distributed intelligence & (distributed intelligence &

IP enabled service creationIP enabled service creation))

Control/Applications DomainControl/Applications Domain(distributed intelligence & (distributed intelligence &

IP enabled service creationIP enabled service creation))MPLS/IP DomainMPLS/IP Domain(switching, aggregation, & (switching, aggregation, &

Layer 2/3 service intelligence)Layer 2/3 service intelligence)

MPLS/IP DomainMPLS/IP Domain(switching, aggregation, & (switching, aggregation, &

Layer 2/3 service intelligence)Layer 2/3 service intelligence)

Qwest

Could optionally add wireless backaul link from a DSLAM to an aggregation point.


5How does 40 Gbps and 100 Gbps help?How does 40 Gbps and 100 Gbps help?

More efficient aggregation of traffic• Today’s large metro, West coast and East coast routes have sufficient demand to Today’s large metro, West coast and East coast routes have sufficient demand to

support higher TDM rates beyond 10 Gbpssupport higher TDM rates beyond 10 Gbps

• Traditionally higher TDM rate easier to manage than DWDM Traditionally higher TDM rate easier to manage than DWDM

Minimizes or eliminates WDM channels • Could be fairly significant in the metro environment as cost avoidance to Could be fairly significant in the metro environment as cost avoidance to

deploying a metro DWDM systemdeploying a metro DWDM system

Reduces the number of customer interfaces to manage

Cheaper – 40 Gbps getting closer• Higher TDM rate interface cards typically prove in at 2.5 times the cost for 4.0 times Higher TDM rate interface cards typically prove in at 2.5 times the cost for 4.0 times

the bandwidth. For example, 10 Gbps proved in at 2.5 times the cost of 2.5 Gbpsthe bandwidth. For example, 10 Gbps proved in at 2.5 times the cost of 2.5 Gbps

Solves customer application ???


6Where are we today ? Qwest has demonstrated Where are we today ? Qwest has demonstrated the largest known 10G/40G BW * D Field Trialthe largest known 10G/40G BW * D Field Trial

Critical achievements:

• ULH DWDM propagation of 85+ 10 Gbps channels at >3000 km with mixed span lengths on Qwest TWC fiber

• 3x40Gbps and 88x10Gbps propagated over 1516 km

1592 1594 1596 1598 1600 1602 1604 1606

-45

-40

-35

-30

-25

-20

wavelength (nm)

po

wer

sp

ectr

um

at

Tx

(dB

m)

10Gb/schannel

40Gb/schannels

10Gb/schann

el

System performance met or exceeded all key advertised features and satisfied Qwest requirements

- Very good performance as measured by both pre- and post-FEC BER

Key caveats• System was based mostly upon GA or System was based mostly upon GA or

near GA equipment and results are near GA equipment and results are realizable in real worldrealizable in real world

• Configuration is not fully optimal and Configuration is not fully optimal and better performance may be achievedbetter performance may be achieved- PMD not known (probably very low –

less than .07ps/sqrt-km)- CD map optimized for 10Gbps (but

sufficient for 40Gbps)• The results do not include tolerances The results do not include tolerances

for best practice engineering (link for best practice engineering (link margins, end-of-life, etc…)margins, end-of-life, etc…)


740G Trial Configuration:1516km40G Trial Configuration:1516km

CHTG T

Seymour

Louisville

Upton

Bowling Green

Nashville

Christiana

Stephenson

3

CHTG(Chattanooga)

TERM

IPLS(Indianapolis)

Seymour

Louisville

Upton

Bowling Green

NashvilleChristianaStephenson

3X3

CHTG T

Seymour

Louisville

Upton

Bowling Green

Nashville

Christiana

Stephenson

3

CHTG(Chattanooga)

TERM

IPLS(Indianapolis)

Seymour

Louisville

Upton

Bowling Green

NashvilleChristianaStephenson

3X3

TERMINALTERMINAL

CHTN Stephenson_TN

Christina_TN

Nashville_TN

BowlingGreen_KY

Upton_KY

Louisville_KY

Seymour_IN

LAS – MA/PA

LAM – MA/DGMA

LAM – DGMA/MA

LAS – PA/MA

LAM – MA/DGMA

LAM – DGMA/MA

LAS – MA/PA

Hardware Loop Backof True Wave Fiber

3X3 SWITCH

MA

MA

MA

IndianapolisTLI Insertion

10 Gig J2127ATransmission Test Set

40GbpsTR

40GbpsTR

40GbpsTR

10GbpsTR

10GbpsTR

100GHz

Mux/Demux

50GH

zTunableG

ateway

10Gbps CWLaser Bank

• 14 tunable 10Gbps (50 GHz) transponders• 3 fixed 40Gbps (100 GHz) transponders• 74 CW lasers


8Where are we going? Key Trends in TransportWhere are we going? Key Trends in Transport

Metro Optical Networking –• Large scale EXC/MSPP with optical interfacesLarge scale EXC/MSPP with optical interfaces• 160Gbps-1.28Tbps capacity160Gbps-1.28Tbps capacity• OC-3c – OC-768(c)OC-3c – OC-768(c)• VCAT combined with L2 capabilitiesVCAT combined with L2 capabilities

Metro Optical Ethernet –• Carrier grade L2 devicesCarrier grade L2 devices• Options for MPLS or VLAN traffic management/organizationOptions for MPLS or VLAN traffic management/organization• Ethernet over dark fiber and over DWDM Ethernet over dark fiber and over DWDM • Fast-E through 10G LAN PHYFast-E through 10G LAN PHY

Long Haul Optical Transport• Expansion of Ethernet in carrier transport space – Ethernet aggregationExpansion of Ethernet in carrier transport space – Ethernet aggregation• Agile optical switching (OXCs, ROADMs)Agile optical switching (OXCs, ROADMs)• Integrated 10G and 40G DWDM systemsIntegrated 10G and 40G DWDM systems• Tunable and integrated optics Tunable and integrated optics • Optical control plane: GMPLSOptical control plane: GMPLS• G.709 digital wrappersG.709 digital wrappers• Ultra FECUltra FEC• Raman amplification options Raman amplification options


9Key Disruptive Optical Trends to WatchKey Disruptive Optical Trends to Watch

Optical component consolidation• Price disruption may drive optical architectures around these optical Price disruption may drive optical architectures around these optical

components. components.

Coherent modulation • May help enable 40 and 100 Gbps bit rates with lower symbol ratesMay help enable 40 and 100 Gbps bit rates with lower symbol rates

Debate over opaque (digital) and transparent (analog) architectures

• Depends on the network physical topology, demand set, applications, Depends on the network physical topology, demand set, applications, etc. etc.

• Some elements of both architectures will be optimalSome elements of both architectures will be optimal


10Optical component consolidationOptical component consolidation

For the last 10 yrs we have been reducing the number of lasers in an optical transport network to reduce cost

• Ultra Long Haul transponders/modulators, FEC, Raman to improve Ultra Long Haul transponders/modulators, FEC, Raman to improve reach and reduce regenerationreach and reduce regeneration

• Transparency to reduce through traffic regenerationTransparency to reduce through traffic regeneration

What if the costs of lasers and optical components became cheap? Different paradigm. Very disruptive.

• Consolidation of multiple discrete optical components onto siliconConsolidation of multiple discrete optical components onto silicon

• Chip yields in full production environment??Chip yields in full production environment??


11Coherent Modulation - Radio Communications 101Coherent Modulation - Radio Communications 101

Today’s optical transport equipment uses OOK modulation• Some version of NRZ, RZ, & CS-RZ formats Some version of NRZ, RZ, & CS-RZ formats

Future coherent modulation is disruptive in the long haul• QPSK, PSK, dQPSK, dPSK, QAM, etcQPSK, PSK, dQPSK, dPSK, QAM, etc

• Offers up to a theoretical gain of 3 dB OSNR over OOK Offers up to a theoretical gain of 3 dB OSNR over OOK

• More complex receiver designMore complex receiver design

• Appears to be more tolerant to some nonlinear effects, chromatic Appears to be more tolerant to some nonlinear effects, chromatic dispersion and PMD dispersion and PMD


12Optical Networking ChoicesOptical Networking Choices

Opaque (Digital) network Transparent (Analog) Network

EXCEXCEXC

TR

SR

Passthru Wavelengths Passthru Wavelengths

EXCEXCEXC

Opaque (Digital)

• Additional grooming pointsAdditional grooming points

• Simpler designSimpler design

• More optical signal More optical signal regeneration on passthru regeneration on passthru wavelengthswavelengths

Transparent (Analog)

• Longer reach, More complex Longer reach, More complex modulation and amplification modulation and amplification systemsystem

• Low cost passthru wavelengthsLow cost passthru wavelengths

• More upfront cost with more all More upfront cost with more all optical components optical components

Answer: For Metro and Long Haul Some of both


1340 Gbps - Challenges and Answers40 Gbps - Challenges and Answers

Fiber• PMD on older single mode fiberPMD on older single mode fiber

Costs• Not quite there Not quite there

• The technology is in-place and emerging with GA products

• Qwest has demonstrated in the field that 40 Gbps is viable over real fiber over varying span lengths in the metro and long haul network

• Good Polarization Mode Dispersion characteristics on TrueWave fiber

• Likely sweet spot for metro transport as a cost avoidance to DWDM

• Coherent modulation schemes will improve long haul economics

Challenges

Answers


14100 Gbps - Challenges and Answers100 Gbps - Challenges and Answers

Fiber• PMD serious problem on today’s fiber, requires PMD compensatorsPMD serious problem on today’s fiber, requires PMD compensators

• Dispersion managed fiber likely required and active dispersion compensatorsDispersion managed fiber likely required and active dispersion compensators

Component and material limitations• Transmitter/Receivers – near electro-optics limits, may require OTDM vs ETDMTransmitter/Receivers – near electro-optics limits, may require OTDM vs ETDM

• Lasers – very narrow, high repetition pulse rate with low jitterLasers – very narrow, high repetition pulse rate with low jitter

• Processors – processing at line rate challenging (limits FEC, for example)Processors – processing at line rate challenging (limits FEC, for example)

System concerns• Channel spacing fairly wide to avoid FWM Channel spacing fairly wide to avoid FWM

• OSNR challenged for current long haul amplifier spacingOSNR challenged for current long haul amplifier spacing

• Nonlinear effects: SPM, XPMNonlinear effects: SPM, XPM

• Not yet clear whether 100 Gbps offers net benefits

• 100 Gbps line rate very tough; may need new fiber for long haul transport

• Coherent modulation and Inverse Muxing appear most promising Allows 100 Gbps to be transmitted as multiple 10/20 Gbps signals

Challenges

Answers

Thank You!Thank [email protected]@qwest.com

Documents

Beyond 10 GbE – Looking Ahead Qwest Communications International Mark Stine, CTO Government Services Division February 2005