How GreenTouch is Making a Difference

Rod TuckerCentre for Energy-Efficient Telecommunications

University of Melbourne

Source: Hilbert and Lopez, “The world’s technological capacity to store, communicate, and compute information,” Science, April 2011

1985 1990 1995 2000 2005 2010 20151E+018

1E+019

1E+020

1E+021

1E+022

1E+023

1E+024

Storage (bits)

Telecommunications (bits/year)

1024

1023

1022

1021

1020

1019

1018Cap

acity

(b

its,

bits

/yea

r)

Global Technology Capacity

Tota

l Po

we

r C

onsu

mp

tion

(W)

Year

2010 2015 2020 2025109

1011

1010

1012

15% p.a. efficiency gains

Global electricity supply

1013

0% p.a. efficiency gains

Power Consumption of

Internet

Sources: Hinton et al., Tucker, IEEE

Power Consumption of the Internet

GreenTouch goal

~3000 kg Co2e/person return

2 X 1 Gb/s for 16 hours ~ 10 TB

~50 kg Co2e/person

Air Travel

Video Conferencing

Melbourne Bangalore

Business Meeting

Travel Replacement

Summary

• Energy implications of network growth– Methods for estimating network energy

• Energy modelling of the network, including data centers

• Key contributors to network energy consumption

• The gap between current practice and “fundamental” limitations– Putting the factor of 1000 into context

• How GreenTouch is making a difference

• Why your organization should join GreenTouch

1. Inventory & sales approach:

Sales (units/year)

Lifetime (years)

Residential

Usage (hours/year)

Power needs

Power management

Equipment stock:

- Residential- Commercial

- Industrial

Equipment consumption:- Residential- Commercial

- Industrial

Energy use:- Residential- Commercial

- Industrial

Total energy

Industrial

Commercial

Total sock

Source: Kawamoto et al. 2001, LBNL-45917

Estimating ICT Energy

2. Transaction-based network modelling approach (GreenTouch):– Calculate power of resources required to deliver services– Network equipment depends on service type– Include network design rules

Service Network path

General Web Fixed access + Metro/Edge (x2) + Long haul + Enterprise

Video Fixed access + Metro/Edge + Long haul + Enterprise

Peer to Peer Fixed access (x2) + Metro/Edge (x2) + Long haul

Mobile data Mobile + Metro/Edge (x2) + Long haul + Enterprise

Mobile voice Mobile (x2) + Metro/Edge + Long haul

Estimating ICT Energy

TDM/WDMring

ADM

EthernetSwitch

Edge Routers

BroadbandNetwork Gateways

ADM

Server

Switch

Gateway

Storage

Fibre

Core Router

WDM

TDM

IPWDM

TDM

IP

WDM

TDM

IP

WDM

TDM

IP

WDM

TDM

IP

Core

Metro/Edge

Base station

SplitterFiber

SwitchFiber

DSLAM

Cu

Data Centre

SplitterFiber

Switch

Fiber

Cu

DSLAM Access

Network Segmentation & Modeling

Metro/Edge

Access Network

• Customer home terminal– ADSL modem, ONU,

wireless/cable modem,..• Access network

field equipment– PON splitter,

DSLAM, RF amps,..• Central office equipment

– OLT, gateway, switch, base station,..

Splitter GPON

PtP

Edge Node

Cabinet

FTTN VDSL2

Cu

Fiber

Fiber

RF Gateway

HFCCu

RF Amp

Node

LTEFiber

Splitter

DSLAM

Switch

Base station

• Ethernet switches and transport• Border Network Gateway (BNG)• Broadband Remote Access Server (BRAS)• Metro transport (TDM, WDM)• Edge routers

Metro/Edge Network

ADM

TDM/WDMring

ADM

EthernetSwitch

Edge Routers

Broadband Network Gateways

Access

EthernetSwitch

Broadband Network Gateways

Edge Routers

Access

Metro/Edge

• Core routers & switches– Number of router hops

• Long haul & submarine optical WDM transport– EDFAs, Raman amps, transmit & receive units, etc

• TDM and WDM cross connects and OADM– Wavelength Selective

Switches, MEMS

Fibre

EDFA

Core Router

WDM

TDM

IP

WDM

TDM

IP

WDM

TDM

IP

WDM

TDM

IP

WDM

TDM

IP

Core Access

Access

WDM

TDM

IP

WDM

TDM

IP

Access

Core Network

• Data centres– Storage disks and arrays– Servers– Local Access Networks– Gateway router– Ethernet switches

• Enterprise networks• Content Distribution Networks

– Centralised– Distributed– Nano Data Centres (Nada)

Servers

Storage

Switches

Gateway routers

Internet

Data Centers and Content Servers

Energy Efficiency of EquipmentE

ner

gy

pe

r bi

t (n

J)

0.01

0.1

1

10

100

MEMSOXC

OpticalAmp

PICTx/Rx

WDM

Tx/Rx

Ethernet

Switch

Core

RouterFEC Chip

0.001

Sub-wavelength Wavelength

2010 Data

Source: Tucker et al., 2009

Source: O. Tamm et al. BLTJ Vol.14, No.4, p.311, 2010

En

erg

y p

er B

itEnergy Efficiency of Equipment20 nJ

10 nJ

1

10

100

1000

10000

1985 1990 1995 2000 2005 2010 2015

nano

-Jou

les

per b

it

Year

Router Energy Efficiency

Cisco AGS

Wellfleet BCN

Cisco GSR 12000

Cisco GSR 12000b

Avici TSR

Cisco CRS 1

Cisco CRS-3

ALU7750

Source: Neilson, 2011

Actual improvementmay be declining

Linear fit gives~25% improvement p.a.

Router Energy Efficiency Trends

En

erg

y/B

it (

nJ)

Transport Energy Efficiency Trends

First Trans-Atlantic

MarconiTrans-Atlantic

FessendenTrans-Atlantic

NY - Paris

Key West - Havana

TAT-1

TAT-3

TAT-5TAT-8

TAT-9TAT-10

TAT-11

TAT-12/13

Newhaven - Azores

10 -

6

1840 1860 1880 1900 1920 1940 1960 1980 2000 2020

Year

En

erg

y/B

it/1

000

km (

mJ)

Wireless

Telegraphy

Coax

Optical + Regen

Optical + EDFA

10 -

4

10 4

10 -

2

10 2

10 6

10 8

1

~15% improvement p.a.Current

Source: Tucker 2011

Summary

• Energy implications of network growth– Methods for estimating network energy

• Energy modelling of the network, including data centers

• Key contributors to network energy consumption

• The gap between current practices and “fundamental” limitations– Putting the factor of 1000 into context

• How GreenTouch is making a difference

• Why your organization needs to join GreenTouch

Access Network Consumption30

Peak Access Rate (Mb/s)

Po

wer

Per

Use

r (W

)

1 10000

FTTP

100

20

10

10

FTTN

Wireless

HFC

FTTP is “greenest”

Global Power Consumption

Year

Tota

l N

etw

ork

Po

wer

Co

nsu

mp

tio

n (

W)

109

1011

1010

108

1012

2010 2015 2020 2025

Total (using 2010 Technology)

Total (15% p.a. efficiency improvements)

Routers and switches

PON

Transport

40% p.a. Access Rate Growth

10% p.a. Growth in user numbers

1.5 billion users

Global electricity supply

Sources: Hinton et al., Tucker, JSTQE , 2011(b)

Year

En

erg

y p

er b

it (m

J)

1.0

100

0.1

10

0.012010 2015 2020 2025

0.1

1

10

100

Ave

rag

e A

cce

ss R

ate

(M

b/s)

Total (using 2010 Technology)

Total (15% p.a. improvements)

PON

Routers and switches

Transport

Energy per Bit

Routers and Switches10-6

10-9

10-10

10-8

10-11

10-7

10-5

10-12

Current Trends

Theoretical Lower Bounds

x 104

Year2010 2015 2020 2025

Ne

two

rk E

ne

rgy

pe

r b

it (

J)

Transport

Switches

Transport

X 102

GreenTouch

X 103

X 102

“Theory” and Practice

Access

Summary

• Energy implications of network growth– Methods for estimating network energy

• Energy modelling of the network, including data centers

• Key contributors to network energy consumption

• The gap between current practices and “fundamental” limitations– Putting the factor of 1000 into context

• How GreenTouch is making a difference

• Why your organization needs to join GreenTouch

GreenTouch Committee Structure

Services & Apps & Trends Committee

Operations Committee

Access Networks Core Networks

Mobile Communications

Fixed Access

Networks

Working Groups

Executive Board

Technical Committee

Network Committee

Core Switching &

Routing

Peter Vetter

Gee RittenhousePeter Winzer

Jaffar Elmirghandi(Tomorrow)

Thierry KleinDavid Neilson (Tomorrow)

Core Transmission

Rouzbeh Razavi

Subrat Kar(Tomorrow)

• Standardized in ~2002• Wildly deploy in last 5 years • GPON : 16 users @ 155Mb/s• EPON : 8 users @ 155Mb/s

Trends in PON

Year2000 2010 2020

100

1000

GPON

Acce

ss-r

ate

(Mb/

s)

EPON

XG-PON10G-EPON

Future

IEEE 10G-EPON and ITU XG-PON• 32 users @ 312Mb/s (20km)• 64 user @155Mb/s (reduced

reach)

Questions• Can PON works at 40Gb/s?• Which architecture should we use?• Which technologies can help reducing

the energy consumption?

ONU 1

TxOLT-MAC

(FPGA) Rx

ONT-FPGA UN

ILine card

DS: 10Gbit/s

ONU 2

DS: Bit-interleaved data

BIPONDS

BIDS

ONU 1

TxOLT-MAC

(FPGA)Rx

ONT-FPGA

UN

I

Line cardDS: 10Gbit/s

ONU 2

DS: Packet data

XGPONDS

XGPONDS

Deser

Deser

Delta

Wireline Access NetworksBit-Interleaved PON (Bi-PON)

Peter Vetter (Tomorrow)

Future 40-Gb/s Access Options

DRIVER

TIA

WC

PD

OLT

OADCF

Active optical splitter

Electronic Switch

DRIVER

TIA

WC

LD/Mod.

PD

OLT

1.5mm

1.3mm

O/E

Proc.

E/O

40 Gb/s Burst

40 Gb/s 1.25 Gb/s

CWDM/TDM PON

WC

lDS1,..,lDS4

lUS1,..,lUS4

lDS1lDS2lDS3lDS4

MU

X

ONU

ONU

ONUlUS1lUS2lUS3lUS4

DE

MU

X

lDS2

lUS2

10 Gb/s Burst

LD/Mod.

DWDM

WC

lDS1,..,lDSN

lUS1,..,lUSN

lDS1

lDSN

MU

X

ONUlUS1

lUSN

DE

MU

X

lDS,i

lUS,i

1.25 Gb/s

1≤i≤N

Source. : Sedighi el al., Paper JTh2A.59 OFC’12

Active Optical Splitter CWDM/TDM

DWDM Electronic Switch

Time (year)2010 2012 2014 2016 2018 2020

2

4

8

16

32

Pow

er c

onsu

mpt

ion

per u

ser (

W)

Active optical splitter (AOS)

CWDM-TDM

DWDM

Electronic Switch (ES)

27 nJ/bit

5.7 nJ/bit

4.6 nJ/bit

1.8 nJ/bit

2.5 nJ/bit

6 nJ/bit

Aggregated rate : 40 Gb/sNumber of ONU : 64

Power and Energy Consumption

• TDM based• > 10 Gb/s ONUs• CMOS improvement is

more severe

• Non-TDM based• <1.25 Gb/s ONUs

Core Networks

......

TDM

IP

WDM ............

TDM

IP

WDM Access

Core Network

Core Optical Networks & Transmission

Core Switching & Routing

TDM

IP

WDM

Core Switching & Routing

Core Switching & Routing

Server

Storage

Data Center

Core Switching & Routing

Acce

ss

Core Switching and RoutingScorpion (silicon photonic interconnect and single-chip

linecard)

Core Switching and RoutingOpera (Optimal end-to-end resource allocation)

Summary

• Energy implications of network growth– Methods for estimating network energy

• Energy modelling of the network, including data centers

• Key contributors to network energy consumption

• The gap between current practices and “fundamental” limitations– Putting the factor of 1000 into context

• How GreenTouch is making a difference

• Why your organization needs to join GreenTouch

VirtualHGW

Un-cooled tunable lasers

Low power OFDM in optical access

Min. energy access architecturesFiber in the Home

Novel PON protocols;Low power CPE

Hybrid PONAlso: TNOZTE, KAIST

PON Sleepmode

Collaboration Among Diverse GroupsExample: Fixed access networks

• BCG2: Beyond Cellular Green Generation*• GTT: Green Transmission Technologies*• LSAS: Large Scale Antenna Systems*• Minimum Energy Access Architectures• VHG: Virtual Home Gateway• OPERA: Optimal End to End Resource Allocation• STAR: Switching & Transmission• REPTILE: Router Power Measurements• Single Chip Linecard• ZeBRA: Zero Buffer Router Architectures• SEASON: Service Energy Aware Sustainable Optical Networks*• HALF MOON: Highly Adaptive Layer for Mesh On-off Optical Networks• EFICOST: Energy Efficient High Capacity OFDM Signal Transmission• Telecommunication Audits and Data Aggregation

Exciting Projects

*Cluster project made up of several sub-projects/activities

An Opportunity to Reinvent the Internet

Conclusions• GreenTouch goal

– Reduce energy per bit by 1000 in a 2020 target network– Requires 100 times more improvement than achievable via Moore’s Law

• Multi-pronged approach– Fixed access– Wireless access– Switching and routing– Core networking– Services and applications

• Great strides have been made – more to come

• Greentouch needs you