Transcript Template - Elsevier

HANDBOOK ON GREEN INFORMATION AND
COMMUNICATION SYSTEMS
Chapter 18:
Towards Energy-Oriented
Telecommunication Networks
1Sergio
Ricciardi, 2Francesco Palmieri, 3Ugo Fiore,
1Davide Careglio, 1Germán Santos-Boada, 1Josep Solé-Pareta
1Technical
University of Catalonia, Spain
2Second University of Naples
3University of Naples Federico II
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
1
Table of contents
Introduction
Background & Motivations
Network energy consumption
Energy-aware architectures
Energy models
Multilevel approach
Energy Efficiency & Energy Awareness
Energy Oriented Architectures
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
2
Network energy consumption
 In Italy, Telecom Italia, and in France, France Telecom, are the second
largest consumers of electricity after the National Railway systems: 2
TWh per year.
 In the UK, British Telecom is the largest single power consumer. [7][8]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
3
The Growing Dynamics
 CPU computing power doubles every 18 months
(Gordon Moore, 1964)
 The available network bandwidth doubles every 6
months (George Gilder, 1992)
Gilder’s
Law
Moore’s Law
4
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
4
Network energy consumption
 Moore’s and Gilder’s laws [2][3] have not had the expected counterpart in power
consumption reduction (Jevons paradox [4])
 As a consequence, the total power required per node is growing faster and faster.
Network infrastructures consume:
 22 GW  1,16% worldwide produced electrical energy
 Growth rate: 12% per year [24]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
5
Network energy consumption
Evolution of the bandwidth capacity and energy-per-bit consumption
Source: [24]
Period windows: 10 years
Bandwidth increment: x1000
Energy per bit: ÷ 100
Energy consumption increment: x10 more than 10 years ago
Technological advancements foreseen by Moore’s law have not been fully compensated by
the same growth in energy-efficiency [47]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
6
Energy consumption: access vs backbone


Backbone
energy
Consumption
Backbone
energy
Consumption
2009:
< 10%
2017:
~ 40%
Energy consumption currently dominated by the access network because of the high number of
end-point devices [31]

ADSL link: 2.8 W, while using as the access infrastructure

GPON (gigabit-capable passive optical network): only 0.5 W (80% improvement)
With rising traffic volume, the major consumption is expected to shift from access to core
networks [12]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
7
Environmental impacts

Human’s activities have severe
impacts on the environment



Human ecological footprint



measures the humanity’s demand on
the biosphere
1,5 planet Earths in 2007 [48]
Carbon footprint


Energy-consumption, resource
exploitation
GHG emissions, climate changes,
global warming & dimming, pollution
Measures the total set of GHG emissions
Three dimensions



Energy consumption (Wh)
GHG emissions (kg CO2)
Energy Cost (€)
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
8
Alternate energy sources


Major contributors to the GH effect:
 water vapor 36–72%
 carbon dioxide CO2 9–26%
 methane 4–9%
 ozone 3–7%
 nitrous oxide ~1%
 chlorofluorocarbons ~1%
Global Warming Potential (GWP)
CO2 equivalent (CO2e)
 Green & dirty energy sources
 Green: solar, wind, tide, geothermic
 Dirty: fossil fueled (coal, oil, gas) and
nuclear plants
9
 Green sources are renewable
 Use Green source as possible!
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
9
Renewable energies
 Advantages:




Virtually endless (5.000 millions years)
Zero carbon (green): no GHG emissions nor pollution during the use phase
Free energy: zero costs in the use phase (except maintenance operations)
Renewable Energy Sources are beneficial over their entire Life-Cycle [34]
 Drawbacks



Low yield (solar panels efficiency ~25% as maximum)
Not always available (e.g., day/night cycle) nor always foreseeable
Not always applicable as the only energy sources (e.g. mobility)
 Allow paradigm shift from centralized to distributed energy system (Smart Grid)


Centralized system: 1 big power plant giving energy to the whole region
Distributed system: n small power plants providing energy for private use and putting surplus
energy into the power grid
 Follow-the-whatever paradigm
 Energy form a cost to a revenue source
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
10
Towards energy-efficiency
 Energy-Efficiency
 refers to a technology designed to
reduce the equipment energy
consumption without affecting the
performance, according to the do
more for less paradigm. It takes
into account the environmental
impact of the used resources and
constraints the computations to be
executed taking into account the
ecological and potentially the
economic impact of the used
resources. Such solutions are
usually referred to as ecofriendly solutions.
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
11
Is energy efficiency sufficient?
 The simplest approach is improving the efficiency of the individual devices
involved
 Energy efficiency alone is not sufficient to achieve effective results (the
Jevons paradox occur)
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
12
The energy-oriented paradigm
 Energy-Awareness
 refers to an intelligent technology that
adapts its behavior or performance
based on the current working load
and on the quantity and quality of
energy that the equipment is expending
(energy-feedback information). It
implies knowledge of the (dirty or
green) sources of energy that supply
the equipment thus differentiating how it
is currently being powered. Energyaware solutions are usually referred to
as eco-aware solutions.
 Energy-Oriented Infrastructures
 Energy-Efficiency + Energy-Awareness
in a holistic, sustainable and
systemic approach with smart grid
power distribution network and
entire life cycle assessment (LCA)
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
13
An energy oriented approach
 Energy as an additional constraint to operate into
network and data centers infrastructures





Current worldwide energy shortages
Rising costs of energy (as fossil sources become scarcer)
Green House effect, Global Warming and Pollution
Need for alternative and renewable energy sources
Growing interests of governments and society into eco-concerns
 Lack of a comprehensive energy-oriented paradigm
 Energy-efficient + energy-aware solutions in a systemic approach
 Renewable energy sources
 Energy models
 The basic Principle
 Do more for less!!!
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
14
Energy-aware architectures
Current router architectures are not energy-aware
average
offered load of
75%
UDP traffic with
different packet
sizes
Source:
[25]
UDP traffic with
medium packet size
with different features
enabled
The difference between
idle and heavily loaded
router vary only of 3%
(about 25 W on 750 W)…
Energy consumption is a function of capacity, not throughput
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
15
Energy-aware architectures
…but…
Power consumption for different installed line
cards configurations of the GSR Cisco Router
Power consumption for different installed
line cards configurations of the 7507Cisco
Router
…power consumption of base system is 50% of full line cards configuration
Focus on energy-aware architectures that can adapt their behavior,
and so, their energy consumption, to the current traffic loads
(advocated both by standardization bodies and governmental
programs and assumed in many literature sources [25])
Source: [25]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
16
Optical Networks are cheaper
 Optical transport (WDM) consumes relatively little energy
 Access network dominates at low rates
 Eliminating the O/E/O converters is not
the only solution
 Network routers dominate at higher rates:




reduce hop count
improve router efficiency (technology)
manage routers better (sleep states)
develop better network architectures using
fewer routers
 manage distribution and replication of
contents
Number of Hops in the Internet
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
[31]
17
Optical vs Electronic devices
Total power consumption vs the aggregated bandwidth for
electronic and optical router technologies
Source: [24]
Optical Cross-Connect node (OXC) with micro-electro-mechanical system (MEMS)
switching logic consumes about 1.2 W per single 10 Gb/s capable interface,
whereas a traditional IP router requires about 237 W per port.
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
18
3R Regeneration,Optical amplifiers &
dispersion compensation
 3R regeneration should be avoided as much as possible in planning,
designing and managing new paths throughout an optical
infrastructure (60 W per channel)
use optical amplifiers, try to avoid 3R as much as possible
 EDFA are more performing (higher gain, lower insertion loss, noise
and cross-talk effects) than SOA but have also higher energy
consumption (respectively 25 W and 3 W) [24]
 Use dispersion compensation fibers (DSF ITU-T G.653, NZ-DSF
ITU-T G.655/656) instead of “simple” single mode fiber (SMF, ITUT G.652) will reduce the dispersion of the optical signal traversing the
fiber and reduce the number of required optical amplifiers [47]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
19
Energy-aware architectures
 Router power consumption:
 Fixed part due for the device to stay on
 Variable part somehow proportional to the traffic load
Power (P)
kWatt
Total power consumption
Variable power consumption

Fixed power
consumption due to the
base system
Load (L)
Idle
Router aggregated bandwidth
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
Tbps
20
Sleep mode
 Putting entire network nodes down when they are not used
 Generic problems






Load balancing
Time consuming
Start-up & configuration problem + peak in power usage
Lifetime (MTBF)
Economic CAPEX & OPEX
Per-interface sleep mode / Adaptive rate / Low Power Idle[39] / STOP-START
Energy proportional computing / Downclocking
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
21
ALR & LPI hybrid concept

Idea: temporarily switching off or downclocking unloaded interfaces and line
cards (per interface sleep mode) to save energy

Adaptive Link Rate (ALR) and Low Power Idle (LPI)
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
22
ALR & LPI

ALR: Native and working link rate
 dynamically modify the link rate according to the real traffic needs

LPI: transmission on single interface is stopped when there is no data to send and quickly resumed when new
packets arrive

in contrast with the continuous IDLE signal used in legacy systems

few microseconds Ts = ~2 s , Tw = ~ 4 s (@10Gbps)

Power consumption in LPI: ~10% of active mode

Power consumption in transition ~50%-100% of active mode

Depends on the packet arrival distribution -> buffering, packet coalescing and coordinated Ethernet
[6][12][13]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
23
A Multi-layer approach
 Network Administration Procedures
 Are aware of the power state of the nodes and the currently set power
policy
 Network Engeneering and Management
 Load balancing, scheduling & task distribution
 Shut down idle nodes
 Networking Protocols




Energy-efficiency, energy-awareness, energy-oriented paradigm
Decrease overall usage of network
Optimized placement/replication of data to minimize equipment usage
Minimize time to transfer data & reduce data to be transmitted (forecasting
techniques)
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
24
Energy-oriented network infrastructure
Source: [47]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
25
Green network control plane
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
26
Conclusions
 Energy as a new constraint – considering the current energy
consumption growth rate – even stronger than the bandwidth capacity
 Energy-aware architectures are necessary to minimize the energy
consumption
 Assess the power consumption of ICT and network infrastructures

In general:
 Energy-efficient architectures
 Energy-aware algorithms & protocols
 Energy-oriented infrastructures (sustainable systemic approach + LCA)
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
27
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
Gartner press release, http://www.gartner.com/it/page.jsp?id=503867, 2007.
An inefficient Truth by the Global Action Plan, http://www.globalactionplan.org.uk/upload/resource/Full-report.pdf.
SMART 2020: Enabling the low carbon economy in the information age, The climate group, 2008.
EU Spring Summit, Brussels, Mar. 2007.
Global Action Plan Report, An inefficient truth, http://www.globalactionplan.org.uk/, 2007
C. Lange, “Energy-related Aspects in Backbone Networks”, in Proc. ECOC 2009, Vienna, Austria, Sep. 2009.
S.Pileri, “Energy and Communication: engine of the human progress”, INTELEC 2007 keynote, Rome, Italy, Sep. 2007.
L. Souchon Foll, “TIC et Énergétique: Techniques d'estimation de consommation sur la hauteur, la structure et
l'évolution de l'impact des TIC en France”, Ph.D. dissertation, Orange Labs/Institut National des Télécommunications,
2009.
BT Press, “BT announces major wind power plans”, Oct. 2007,
http://www.btplc.com/News/Articles/Showarticle.cfm?ArticleID=dd615e9c-71ad-4daa-951a-55651baae5bb.
Telefónica supplement, The environment and climate change, 2008 special report on corporate responsibility, Apr. 2009.
H.D. Saunders, “The Khazzoom-Brookes postulate and neoclassical growth”, The Energy Journal, Oct. 1992.
C. Lange, D. Kosiankowski, C. Gerlach, F. Westphal, A. Gladisch, “Energy Consumption of Telecommunication
Networks”, in Proc. ECOC 2009, Vienna, Austria, Sep. 2009.
Nature Photonics technology conference 2007, Tokyo, Japan, Oct. 2007.
M. Gupta, S. Singh, “Greening of the Internet”, in Proc. ACM SIGCOMM 2003, Karlsruhe, Germany, Aug. 2003.
R.S. Tucker et al., “Evolution of WDM Optical IP Networks: A Cost and Energy Perspective”, IEEE/OSA Journal of
Lightwave Technologies, vol. 27, no. 3, pp. 243-252, Feb. 2009.
B. St Arnaud, “ICT and Global Warming: Opportunities for Innovation and Economic Growth”,
http://docs.google.com/Doc?id=dgbgjrct\2767dxpbdvcf.
A. Muhammad, Paolo Monti, Isabella Cerutti, Lena Wosinska, Piero Castoldi, Anna Tzanakaki, “Energy-Efficient WDM
Network Planning with Protection Resources in Sleep Mode”, accepted for Globecom 2010, ONS01.
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
28
References
18. L. Chiaraviglio, M. Mellia, F. Neri, “Energy-aware Backbone Networks: a Case Study”, GreenComm – First International
Workshop on Green Communications, Dresden, Germany, Jun. 2009.
19. W. Van Heddeghem, M. De Groote, W. Vereecken, D. Colle, M. Pickavet, P. Demeester, “Energy-Efficiency in
Telecommunications Networks: Link-by-Link versus End-to-End Grooming”, in Proc. of ONDM 2010, Feb. 1-3 2010,
Kyoto, Japan.
20. R. S. Tucker, “Modelling Energy Consumption in IP Networks”, retrieved from:
http://www.cisco.com/web/about/ac50/ac207/crc_new/events/assets/cgrs_energy_consumption_ip.pdf.
21. W. Vereecken, W. Van Heddeghem, D. Colle, M. Pickavet, P. Demeester, “Overall ICT footprint and green
communication technologies”, in Proc. of ISCCSP 2010, Limassol, Cyprus, Mar. 2010.
22. Juniper, http://www.juniper.net.
23. M. Z. Feng, K. Hilton, R. Ayre, R. Tucker, “Reducing NGN Energy Consumption with IP/SDH/WDM”, in Proc. 1st
International Conference on Energy-Efficient Computing and Networking, Passau, Germany, pp: 187-190, 2010,
ISBN:978-1-4503-0042-1.
24. BONE project, 2009, “WP 21 Topical Project Green Optical Networks: Report on year 1 and updated plan for activities”,
NoE, FP7-ICT-2007-1 216863 BONE project, Dec. 2009.
25. J. Chabarek, J. Sommers, P. Barford, C. Estan, D. Tsiang, and S. Wright, “Power awareness in network design and
routing”, in Proc. IEEE INFOCOM, 2008.
26. S. Aleksic, “Analysis of Power Consumption in Future High-Capacity Network Nodes”, Journal of Optical
Communications and Networking, vol. 1, no. 3, pp. 245-258, Aug. 2009.
27. I. Cerutti, M. Tacca, A. Fumagalli, “The multi-hop multi-rate wavelength division multiplexing ring”, IEEE/OSA Journal of
Lightwave Technologies, vol. 18, 2000.
28. Energy Star, “Small network equipment”, http://www.energystar.gov/index.cfm?c=new_specs.small_network_equip.
29. Gordon E. Moore, “Cramming more components onto integrated circuits”, Electronics, Volume 38, Number 8, April 19,
1965.
30. George F. Gilder, “Telecosm: How Infinite Bandwidth Will Revolutionize Our World”, The Free Press, NY, 2000.
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
29
Thanks for your
attention!
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II
30