Energy Consumption in Energy Consumption in IP NetworksIP Networks

Rodney S. Tucker, Jayant Baliga, Robert Ayre, Kerry Hinton, Wayne V. Sorin

ARC Special Research Centre for Ultra-Broadband Information Networks (CUBIN)

University of Melbourne

r.tucker@ee.unimelb.edu.au

Energy Consumption of the NetworkEnergy Consumption of the Network

• Greenhouse Impact• Managing “Hot Spots”

- Getting the energy in- Getting the heat out

• Energy-limited capacity bottlenecks

Why should we be interested in energy?

• OPEX

Hot spot

• Enabling energy efficiencies in other sectors

Power In

Energy Consumption GrowsEnergy Consumption Grows

More users

More data-intensive applications, e.g. video

More often and for longer periods

Increasing demand → operators provide faster access and increased core capacity

New applications enabled by faster access

Where are We Heading ?Where are We Heading ?

SummarySummary

Modeling energy consumption of the Internet- Core, metro, and access networks

Where does the energy go?

Will (can) optical switching technologies help to reduce energy consumption?

A Word of Warning

What is the Carbon Footprint of Telecoms?

Adapted from “SMART 2020: Enabling the low carbon economy in the information age,” GeSI, 2008 www.gesi.org

Global Telecoms Footprint (devices & infrastructure)

Mobile Network

1450+% growth

0 100 200 300 400

2020

2002

Footprint (MtCO2 p.a.)

Fixed Narrowband

Broadband Modems

Fixed Broadband

Mobile handsets

20X increase

Metro

Core

Edge Edge

Curb Curb Curb Curb

Core Core

Access

Core

ONU ~ 5-10W

OLT - 100W

12816 Edge ~ 4 kW

CRS-1 ~ 10 kW / rack

0.1 - 1000 Mb/s to the user

Fibre Amps

WDM

Passive Optical Network

Packet over

Sonet

Where does the Energy Go? Where does the Energy Go?

Baliga et al., 2007

Core Core

Core

OLT - 100W

Number of Hops in the InternetNumber of Hops in the Internet

0 5 10 15 20 250

0.02

0.04

0.06

0.08

0.1

Number of Hops, k

[]

Pr

Hk

=

Source: P. Van Mieghem,“Performance Analysis of Computer Systems and Networks”, Cambridge (2006)

2006 Data

Multi ISP Network: 15 - 20 hops

25000

Pow

er (W

/use

r)

% o

f Ele

ctric

ity S

uppl

y

Baliga et al., 2007

0

20

0.5

25

5

Peak Access Rate (Mb/s)

15

100 20015050

1.0

10

Power Consumption of IP NetworkPower Consumption of IP Network

Total

Routers

Access (PON)

SDH/WDM Links

Today’s Internet (~ 2.5 Mb/s)

2007 Technology20 router hops

Contention ratio = 25

Pow

er (W

/use

r)

% o

f Ele

ctric

ity S

uppl

y

Baliga et al., 2007

Total

UltraUltra--Broadband AccessBroadband Access

2007 Technology

Peak Access Rate (Mb/s)

00

2.550

100 5.0

Routers

Access (PON) SDH/WDM

0 1000500

20 router hops

Contention ratio = 25

Energy Consumption in Access NetworksEnergy Consumption in Access NetworksNEC CM7710T

Splitter PON

PtP

WiMAX

Cabinet

Edge Node

Cisco 12816

NEC CM7700S

ZyxelVES-1616F-34Cisco

4503

NEC VF200F6

NEC GM100

Axxcelera ExcelMax CPE

AxxceleraExcelMax BTS

Cabinet

Access N/W

FTTN with VDSL2

Pow

er P

er U

ser (

W)

Peak Access Rate (Mb/s)1 100

WiMAX

FTTN

0

40

PtP

PON

10 250

Oversubscription = 10

Power Consumption in Access NetworksPower Consumption in Access Networks

• Wireless access consumes more energy than optical access• PON FTTH is “greener” than FTTN

25

20

Total

Access (PON)Routers

2.5 25 250 2500

Ener

gy p

er b

it (J

)

10-6

10-3

10-8

10-5

10-7

10-4

20 hops

~1 μJ/b

WDM Links

Peak Access Rate (Mb/s)

Network Network EnergyEnergy Consumption per BitConsumption per Bit

~100 μJ/b

• Optical transport (WDM) consumes relatively little energy< 5% of energy > 25% of CAPEX

• Eliminating the O/E/O converters provides no significant benefit

• Access network dominates at low rates– Standby/Sleep mode is key to reducing energy consumption

• Network routers dominate at higher rates– Need to

• reduce hop count• improve router efficiency (technology)• manage routers better (sleep states)• develop better network architectures using fewer routers• manage distribution and replication of content (IPTV)

ObservationsObservations

Router Throughput

Pow

er c

onsu

mpt

ion

(W)

Source: METI, 2006, Nordman, 2007

Power Consumption in RoutersPower Consumption in Routers

10

100

1,000

10,000

100,000

1,000,000

P = C2/3

where P is in Wattswhere C is in Mb/s

10 nJ/bit

100 nJ/bit

11 Pb/s

P ~ 10

1 Tb/s1 Gb/s1 Mb/s

?

High-end router: Cisco CRS-1

Linecard ChassisCapacity: 0.64 Tb/s Power: 13.6 kW

Switch Fabric Chassis: Power: 8 kW

Fully equipped:Multi-rack routerCapacity: 41 Tb/sPower ~ 1 MW

Source: Neilsen, 2006; Deutche Telekom, 2007

Per Rack

X2 every18 months

Energy BottleneckEnergy Bottleneck

Optics

Switch Fabrics Buffers

Demutiplexers

Multiplexers

Fibers

Forwarding Engine

J

SwitchFabric

O/E Converters

Reduced bit rate (i.e. parallel processing)

Electronic RoutersElectronic Routers

Speed (throughput) is not a limitation

Electronics

Line Card

Energy in Electronic and Optical RoutersEnergy in Electronic and Optical Routers

G. Epps, Cisco, 2007, ITRS, 2005, R. Tucker, JLT, 2006

BufferI/O

Data Plane

Control Plane

Routing Engine

Routing Tables

Power supply

inefficiency

Fans and blowersO/E

O/E Forwarding Engine

Forwarding Engine

Energy/bit

0.7 nJ 1.0 nJ0.5 nJ3.2 nJ 3.5 nJ1.1 nJElectronic (2008)

Optical Packet Switching is not a compelling alternative

Buffer

Switch Fabric

Buffer

10 nJ

Total

Electronic (2018) 10 pJ 20 pJ10 pJ65 pJ 80 pJ25 pJ 210 pJ

Switch Control

Optical (2018) 20 pJ 20 pJ20 pJ65 pJ 80 pJ25 pJ 220 pJ?

Amp

Amp

Contention Resolution in the Wavelength Domain Contention Resolution in the Wavelength Domain

Switch Fabric

Forwarding Engine

Forwarding Engine

Forwarding Engine

1λ

nλ

Fatal Flaw: Require large n for low blocking probability (n ~3 -10 x)

Pow

er (W

/use

r)

Total (Conventional router)

Peak Access Rate (Mb/s)

0

50

100

Routers

AccessWDM

0 1000500

WDM 5 X

Routers 1.2 X

Total (Wavelength-domain contention resolution )

Wong, JLT 2006 Pathiban et al., JLT 2009

• Optical Burst Switch

• Optical Label Switch

Efficiency Improvement Rate = 0% p.a

5% p.a

Peak Access Rate (Mb/s)1 100 200 300 400

Tota

l Pow

er P

er U

ser (

W)

0

40

60

20

0

% o

f Ele

ctric

ity C

onsu

mpt

ion

1.0

2.0

3.0

10% p.a

20% p.a

Baliga, et al, 2008, unpublished

The ChallengeThe Challenge

10 % -

20 % p.a. continuous improvement in efficiency

Target

SummarySummary

• Energy consumption currently dominated by the access network

• The energy bottleneck in routers is looming- More significant than the so-called “electronic speed bottleneck”

• Key strategies for efficient network design - Control energy in the access network (e.g. sleep mode in modems)- Reduce the hop count (i.e. “agile” optical bypass)- Energy-efficient network architectures- Continuous improvement in (electronic) router efficiency- Caching and content distribution networks- O/E/O conversions not necessarily a problem