2001 Altamar Networks. All rights reserved Designing US Scale Networks By Paul Tomlinson.

2001 Altamar Networks. All rights reserved

Designing US Scale Networks

By Paul Tomlinson

Overview

• Transmission technologies– 4 technologies

• Long Reach & Ultra Long Reach• With & without Raman

– Varying reaches with fibre different types

• Links– Arbitrary lengths– Arbitrary dB loss

• Solution test all design possibilities even the use of technology ‘patch works’

X

Transmission Technology Choice

2,500 km

Hypothetical data, do not use out of example context

Long Reach (1,000km)

Ultra Long Reach (2,000km)

Long Reach + Raman (1,500km)

Ultra Long Reach + Raman (3,000km)

3R 3R3R 3R3R X3R

Good fit Good fit but includes cost of 2 Regens! Regen Required - high cost! Wastage of reach distance & inclusion of 3R cost Mixture of technologies – Case #1Mixture of technologies – Case #2

X

Transmission Technology Choice

2,500 km

Hypothetical data, do not use out of example context

Long Reach (1,000km)

Ultra Long Reach (2,000km)

Long Reach + Raman (1,500km)

Ultra Long Reach + Raman (3,000km)

X

• Cost of transponders• Cost of regenerators• Cost of Raman amps• Number of wavelengths• Fibre, floor space, …

Which technology gives the least cost fit?

Dependencies:

Transmission Technology Choice

Long Reach (1,000km)

Ultra Long Reach (2,000km)

Long Reach + Raman (1,500km)

Ultra Long Reach + Raman (3,000km)

Long Reach only $24m

Ultra Long Reach only $27m

Long Reach + Raman only $27m

Ultra Long Reach + Raman only $15m

Ultra Long Reach & Long Reach + Raman $22m

Long Reach & Long Reach + Raman $19m

Cheapest because no regenerators are required

However this can all change with varying distances, number of wavelengths and hardware costs!

Hypothetical data, do not use out of example context

• 2500km link• 80km amp spacing• Including dispersion compensation• 40 • Representative costs!

Example: least cost transmission technology

Link Distance (km)

Wav

elen

gth

Lo

adin

g

LRULRULR with RamanLR with Raman

Hypothetical data, do not use out of example context

The plot shows the least cost technology for varying link distances and wavelength loadings. It varies for different relative component costs.

Geography and Topology Constraints

Throughout a real network large span distances and poor fiber can reduce dB loss, thus making the minimal cost solution less straight forward

xxNode A Node Z

Large span distances force the need for regeneration

R

{ {

xxNode A Node Z

Or

Raman pump over these difficult spans

But trade-offs exist!

So each possible design must be tested to gain the minimal cost

Geography and Topology Constraints

Throughout a real network large span distances and poor fiber can reduce dB loss, thus making the minimal cost solution less straight forward

xxNode A Node Z

OADM’s reduce transparency distances

OADM

xxNode A Node Z

So

Raman pump to recover the range

But eventually regeneration is needed!

So again, each possible design must be tested to gain the minimal cost

OADMR

OADM OADM

Integrated transmission & switching

Mux/demux+ grooming

Service interfaces

Patch panelShort reachinterfaces

Trunk interfaces

Mux/demux+ grooming

Service interfaces

Trunk interfaces

WDM

WDM

• The following slides were used for the WorldCom presentation – they were based on a generic network design and could be adapted …

Value of switch-transport integration

• 25% capex savings on whole network• Up to 46% capex savings on individual nodes

– Fewer network elements to manage

OC-192 LR/ULR Transponders

OC-192 LR/ULRTransponders

OC-192 LR/ULRTransponders

OC-192 SRTransponders

x

x

IntegratedTitanium

Stand- Alone

Integrated Vs Stand-alone Switch & Transport

Stand-Alone Switch + Transport Integrated Solution

Pri

ce

($

)

Value of integrated OADM

• Fewer transponders & smaller switch matrix reduces capex by further ~13%– Integrated management of pass through channels

xOC-192 LR/ULR

Transpondersall channels

Titanium o-e-o

x

OADM

OC-192 LR/ULRTransponders

Add-drop channels only

Titanium OADM Price Difference with & without OADMs

Solution without OADMs Solution with OADMs

Pri

ce (

$)

Common Equipment

OADMs

Optical Shelves

Switch Shelves

Transponder Shelves

Linear scalability

• Cost scales linearly with growth– All demands were increased by 20% period on period

Incremental Cost Increase

Period 1 Period 2 Period 3 Period 4

Pri

ce (

$)

Spares … need more work

Challenges in designing transport networks

Where should we use OADM’s?Where should we use Raman?

What is the best protection type for a given application?

How do we trade off transparency with connectivity when applying OADM’s?

How can we optimize customer network designs & minimize costs?

Where should we groom?

How valuable is integration?How scalable is

the solution?

How can we compare solutions?

Generic network

• How do opaque and transparent networks compare?– Transparent exploits transparency range but analogue

performance constrains flexibility– Opaque allows grooming and maximises fill