© technical university of catalonia - second university of naples - university of naples federico...
TRANSCRIPT
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 1
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, Spain2Second University of Naples
3University of Naples Federico II
HANDBOOK ON GREEN INFORMATION AND COMMUNICATION SYSTEMS
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 2
Table of contentsTable of contents
IntroductionBackground & MotivationsNetwork energy consumptionEnergy-aware architecturesEnergy modelsMultilevel approachEnergy Efficiency & Energy AwarenessEnergy Oriented Architectures
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 3
Network energy consumptionNetwork 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 4
4
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’sLaw
Moore’s Law
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 5
Network energy consumptionNetwork 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 6
Network energy consumptionNetwork energy consumptionEvolution of the bandwidth capacity and energy-per-bit consumption
Period windows: 10 yearsBandwidth increment: x1000
Energy per bit: ÷ 100Energy 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]
Source: [24]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 7
Energy consumption: access vs backbone
Backbone energy
Consumption
2009:< 10%
Backbone energy
Consumption
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 8
Environmental impacts Human’s activities have severe
impacts on the environment Energy-consumption, resource
exploitation GHG emissions, climate changes,
global warming & dimming, pollution
Human ecological footprint measures the humanity’s demand on
the biosphere 1,5 planet Earths in 2007 [48]
Carbon footprint 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 9
9
Alternate energy sourcesAlternate energy sources
Green & dirty energy sources Green: solar, wind, tide, geothermic Dirty: fossil fueled (coal, oil, gas) and
nuclear plants
Green sources are renewable Use Green source as possible!
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)
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 10
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
Renewable energies
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 11
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 eco-friendly solutions.
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 12
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 13
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. Energy-aware 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 14
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 15
Energy-aware architecturesCurrent router architectures are not energy-aware
Source: [25]
UDP traffic with different packet
sizes
UDP traffic with medium packet size
with different features enabled
average offered load of
75%
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 throughputEnergy consumption is a function of capacity, not throughput
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 16
Energy-aware architecturesEnergy-aware architectures
…power consumption of base system is 50% of full line cards configuration
Source: [25]
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
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])
…but…
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 17
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
[31]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 18
Optical vs Electronic devices
Total power consumption vs the aggregated bandwidth for electronic and optical router technologies
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.
Source: [24]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 19
3R Regeneration,Optical amplifiers 3R Regeneration,Optical amplifiers & dispersion compensation& 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, ITU-T 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 20
Router power consumption:
Fixed part due for the device to stay on
Variable part somehow proportional to the traffic load
Energy-aware architecturesEnergy-aware architectures
Power (P)kWatt
Router aggregated bandwidthLoad (L)
Tbps
Fixed power consumption due to the
base system
Total power consumption
Variable power consumption
Idle
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 21
Sleep modeSleep 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 22
ALR & LPI hybrid conceptALR & 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 23
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 24
A Multi-layer approachA 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 25
Energy-oriented network Energy-oriented network infrastructureinfrastructure
Source: [47]
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 26
Green network control plane
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 27
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 architecturesEnergy-efficient architectures Energy-aware algorithms & protocolsEnergy-aware algorithms & protocols Energy-oriented infrastructures (Energy-oriented infrastructures (sustainable systemic approach + LCA)sustainable systemic approach + LCA)
© Technical University of Catalonia - Second University of Naples - University of Naples Federico II 28
References1. Gartner press release, http://www.gartner.com/it/page.jsp?id=503867, 2007.
2. An inefficient Truth by the Global Action Plan, http://www.globalactionplan.org.uk/upload/resource/Full-report.pdf.
3. SMART 2020: Enabling the low carbon economy in the information age, The climate group, 2008.
4. EU Spring Summit, Brussels, Mar. 2007.
5. Global Action Plan Report, An inefficient truth, http://www.globalactionplan.org.uk/, 2007
6. C. Lange, “Energy-related Aspects in Backbone Networks”, in Proc. ECOC 2009, Vienna, Austria, Sep. 2009.
7. S.Pileri, “Energy and Communication: engine of the human progress”, INTELEC 2007 keynote, Rome, Italy, Sep. 2007.
8. 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.
9. BT Press, “BT announces major wind power plans”, Oct. 2007, http://www.btplc.com/News/Articles/Showarticle.cfm?ArticleID=dd615e9c-71ad-4daa-951a-55651baae5bb.
10. Telefónica supplement, The environment and climate change, 2008 special report on corporate responsibility, Apr. 2009.
11. H.D. Saunders, “The Khazzoom-Brookes postulate and neoclassical growth”, The Energy Journal, Oct. 1992.
12. C. Lange, D. Kosiankowski, C. Gerlach, F. Westphal, A. Gladisch, “Energy Consumption of Telecommunication Networks”, in Proc. ECOC 2009, Vienna, Austria, Sep. 2009.
13. Nature Photonics technology conference 2007, Tokyo, Japan, Oct. 2007.
14. M. Gupta, S. Singh, “Greening of the Internet”, in Proc. ACM SIGCOMM 2003, Karlsruhe, Germany, Aug. 2003.
15. 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.
16. B. St Arnaud, “ICT and Global Warming: Opportunities for Innovation and Economic Growth”, http://docs.google.com/Doc?id=dgbgjrct\2767dxpbdvcf.
17. 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 29
References18. 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 30
Thanks for your attention!