Evolving Metro/Regional ROADM

Architectures to 400G and beyond

ECOC Market Watch

London

September 2013

Paul Momtahan

Director of Transport Solution Marketing

Evolving Metro/Regional to 400G Agenda

Metro/Regional Market Drivers

Moving beyond 100G

ROADM Architecture

Evolution

Cost Effective CDC for the Metro

Metro/Regional Market Drivers Traffic Growth & 100G

Worldwide Annual Growth in Traffic Volume on Metro Networks

% of Respondents

Operator Timelines for Wide-Scale Metro 100G Deployment

% of Respondents

Source: Metro Packet-Optical Transport 2.0 Survey, Heavy Reading, March 2013

Traffic growth is driving operators to start adopting 100G in the metro

Metro/Regional Market Drivers Inter-Data Centre Traffic

Inter-data centre traffic is growing at >30% CAGR

32

44 45

57 62

0

10

20

30

40

50

60

70

Amsterdam Frankfurt Paris NYC London

# o

f C

olo

ca

tion D

ata

centr

es

Inter-data centre traffic is likely to drive the need for higher speed lambdas in the metro

Metro/Regional Market Drivers Metro vs. Long Haul

Local

Exchange

Hub Office

Metro / Regional

Long Haul

Consumer: Hubbed & Asymmetrical

Enterprise: Some meshed, more symmetrical

nX 10Gs

Short distances

Few OLAs

Traditional Metro Consumer/Enterprise

Meshed

Symmetrical Traffic

nX 100/400Gs

Long Distances

Numerous OLAs

Core Network Hub to Hub

Meshed

Symmetrical

nX 100/400Gs

Short distances

Few OLAs

New Metro Layer Data Centre to Data Centre

Will have 10G, 100G and 400G

Focus on Flexibility and Cost

Coherent Only

Performance & Flexibility

Moving beyond 100G 16QAM & Superchannels

16QAM doubles spectral efficiency but is at least four times more sensitive to OSNR

– Technology is limited to 30Gbaud today, thus requiring 2x200Gbps for 400G

16QAM

4x100Gb

137.5 GHz

2x200Gb

75 GHz

1x400Gb

75 GHz 100

Gb

50 G

Hz

100

Gb

50 G

Hz

100

Gb

50 G

Hz

Superchannels & Flexgrid

Superchannels and flexgrid enable a further 33% boost in spectral efficiency by combing two

200Gbps lambdas in 75Ghz

Moving beyond 100G Impact of 10G on 16QAM

The metro is still dominated by 10G however 16QAM is severely impacted by crosstalk (i.e. XPM) from 10G

0 2 4 6 8 10

With 10G & DCMs

Coherent Only

# of 40km/10dB spans (ROADM cascades)

400G (2x200G) reach with 3dB margin

ROADM Architecture Evolution Broadcast & Select

SPLITTER

SPLITTER

SPLITTER

SPLITTER

SPLITTER

Broadcast & Select has been the dominant architecture for multi-degree ROADM

WSS

WSS

WSS

WSS

WSS

As the number of degrees

increases so does the

splitter loss, lowering OSNR

WSS Isolation:

Non-selected channels leak

through and create cross talk

ROADM Architecture Evolution Route & Select

Route & Select does not suffer from splitter scaling or WSS isolation problem

WSS

WSS

WSS

WSS

WSS

WSS

WSS

WSS

WSS

WSS

ROADM Architecture Evolution 400G Reach: Route & Select vs. Broadcast & Select

If we get rid of 10G and Dispersion Compensation, Broadcast and Select can

deliver 400G performance comparable with Route & Select

However as the number of degrees (size of splitter & WSS) increases Broadcast &

Select performance will degrade

Amplifier OSNR Broadcast & Select Isolation (20D ROADM)

Route and Select

R&S has better 400G performance as # of degree scales

16

ROADM Architecture Evolution Broadcast & Select vs. Route & Select Broadcast & Select Advantages

– Cost-effective (1xWSS per Degree)

– Filter narrowing (1xWSS per pass-through) - No longer an issue with next gen WSS

– Support for optical Broadcast/Multicast schemes

– Good 400G performance if no 10G/DCMs and degree count is not too high

Route & Select Advantages

– No loss in splitter (advantage for 400G & high-degree ROADM)

– No leakage from non-selected channels at egress (advantage for 16QAM/400G)

– Scales to support high degree counts (mesh, CD/CDC) without impacting 400G performance

00.20.40.60.8

11.21.41.61.8

B&S, no CD R&S with CD R&S with CDC

Recent Tellabs Study on North American metro

Rela

tive C

ost of O

ptical Layer

Add/Drop Number of Degrees (DWDM Line)

Directionless Colourless Contention-less 2D 4D 6D 8D 10D >10D

ROADM Architecture Evolution ROADM Decision Criteria for 400G

Drivers for colourless & directionless:

– Layer 0 Restoration

– Network Defragmentation/Optimisation (SDN)

– Fast Provisioning (pre-installed transponders)

Drivers for contention-less:

– Minimises blocking in highly loaded network

– Network blocking not node contention is likely to be the main source of blocking in metro networks

Drivers for more DWDM line degrees

– Meshing in the metro, extending to access

– C&D (each directionless eats a degree)

– CDC (max 16 add/drop ports per “degree”)

Add/Drop Number of Degrees (DWDM Line)

Directionless Colourless Contention-less 2D 4D 6D 8D 10D >10D

Add/Drop Number of Degrees (DWDM Line)

Directionless Colourless Contention-less 2D 4D 6D 8D 10D >10D

Broadcast & Select

Route & Select

Cost Effective CDC for the Metro High Port Count MCS

MCS MCS MCS

16 Add/drops

degree 1 degree 2 degree 8

Scalable

– 16 add/drops per WSS port

High loss across MCS

– EDFA array required

– 2xEDFA for each degree (input & output)

– 16xEDFA for eight degree system

Decreased Performance

– EDFAs add noise

– Mixing of 16 sources (SMSR)

Cost Effective CDC for the Metro Low Port Count MCS

MCS MCS MCS

4~8 Add/drops

degree 1 degree 2 degree 8

Cost per add/drop port drops by 30%~50%

– Reduced loss across MCS

– Enables EDFAs to be eliminated

Better performance:

– No noise from EDFA

– Mixing of fewer add/drop sources

Less scalable

– Eats up WSS ports more quickly

Increased WSS cost?

Requires more expensive

high port count WSS for DWDM Line

Cost Effective CDC for the Metro Tellabs Metro CDC Study

LPC MCS is more cost-effective than HPC MCS unless # of wavelengths

increases dramatically

– 100G and 400G are likely to depress the number of wavelengths

CDC with LPC MCS can even be lower cost than CD only (with R&S)

Scale of demands (1=current wavelengths)

Evolving the Metro to 400G Summary

Hub Office

Metro / Regional

• Stick with 10G friendly solution

• Stick with B&S

• Focus on low cost

Traditional Metro Consumer/Enterprise Delivery

• Deploy Overlay 100G/400G friendly

optical Layer

• All Coherent

• Choose between B&S for cost

effectiveness and R&S for flexibility

• LPC MCS is likely to be the most cost-

effective option for CDC – could even be

less expensive than CD only

New Metro Data Centre to Data Centre

Cloud & Virtualization have resulted in an explosion of horizontal

traffic between data centres, creating two completely separate

traffic profiles in metro networks