Charting a Path to Sustainable and Scalable ICT Networks : Rod Tucker, Rob Ayre, Kerry Hinton Centre for Energy-Efficient Telecommunications University of Melbourne
Power Consumption (W) Power Consumption of the Global Internet 10 13 Global electricity supply 10 12 10 11 Power Consumption of Internet (Including servers) 10 10 15% p.a. Improvement in all technology 40% p.a. Data growth 10% p.a. Growth in user numbers 10 9 2010 2015 2020 1.5 billion users Year
Inside the Network Core Network Metro/Edge Network Access Network Core Router Fiber Broadband Network Gateways Ethernet Switch DSLAM Cu DSL Edge Routers OLT PON Splitter Fiber FTTP Hot spots OLT ONU Cabinet DSLAM Cu FTTN Server Storage Data Center Server Storage Content Distribution Network
Access Network Customer home terminal ADSL modem, ONU, wireless/cable modem,. Access network field equipment Edge Node Switch Splitter Splitter Cabinet DSLAM Fiber Cu GPON FTTN VDSL2 PON splitter, DSLAM, RF amps,.. Central office equipment Base station Fiber Fiber PtP LTE WiMAX OLT, gateway, switch, base station,.. RF Gateway Node Cu RF Amp HFC
Data Centers and Content Servers Load-Balancing Switches Border Routers 15% of traffic to users 5% of traffic to other data centers Aggregation Switches 80% of traffic stays in data center Racktop Switches Racks of Servers
Energy per Bit (nj) Energy Efficiency of Key Equipment 20 18 16 14 12 10 8 Source: O. Tamm et al. 2010 Router Packet switch SDH Cross-connect OTN Cross-connect 6 4 2 0 2005 2007 2009 2011 2013 2015 2017 2019 Year
nano-joules per bit Router Energy Consumption Trends 10000 Router Energy Efficiency 1000 Cisco AGS Linear fit gives ~25% improvement pa 100 Wellfleet BCN Cisco GSR 12000 Actual improvement may be declining 10 Cisco GSR 12000b Avici TSR Cisco CRS 1 ALU7750 Cisco CRS-3 1 1985 1990 1995 2000 2005 2010 2015 Year Source: Nielsen, ECOC 2011
Energy/Bit/1000 km (mj) Transport Energy Consumption Trends 10 8 10 6 10 4 10 2 1 10-2 10-4 10-6 First Trans- Atlantic Marconi Trans-Atlantic Fessenden Trans-Atlantic Newhaven - Azores NY - Paris Key West - Havana ~15% improvement p.a. TAT-1 TAT-5 TAT-3 TAT-9 TAT-11 TAT-12/13 TAT-8 TAT-10 1840 1860 1880 1900 1920 1940 1960 1980 2000 2020 Year Wireless Telegraphy Coax Optical + Regen Optical + EDFA Source: Tucker 2011
Compound Average Growth Rate (%) Internet Traffic Growth Trends Source: Kilper et al., JSTQE 2011
Exabytes Exabytes/annum /annum Cisco Projections of Data Centre Traffic 5000 4000 3000 2000 1000 Data Centre Traffic 2010-2015 Data Center to User Between Data Centers Within Data Centers Datacentre to user Between Datacentres Within Datacentres 0 2010 2011 2012 2013 2014 2015 Year Source : Cisco Cloud Index 2011
Power Per User (W) Access Network Energy Consumption 30 Wireless 20 FTTN 10 HFC FTTP (PON) 0 PON is greenest 1 10 100 1000 Peak Access Rate (Mb/s) Source: Baliga et al., OFC 2009
Power Consumption (W) Power Consumption of the Global Internet 10 12 Global electricity supply (3% p.a.) 10 11 Total (2010 Technology) Total 10 10 Access (PON) 10 9 10 8 15% p.a. technology improvement 2010 2015 2020 Year
Energy per User bit (mj) Average Access Rate (Mb/s) Network Energy per User Bit 100 100 10 1.0 Core and Metro Networking PON Total 10 0.1 Optical Transport 1 0.01 0.1 2010 2015 2020 Year
Network Energy per bit (J) Gap Between Theory and Practice 10-5 10-6 X 10 3 Wireline Access Current Trends Routers and Switches 10-7 10-8 Transport X 10 2 10-9 10-10 X Switches Lower Bounds X 10 2 x 10 4 10-11 Transport X X 10-12 2010 2015 2020 2025 Year Source: Tucker, JSTQE 2011
A Path to Sustainable ICT Networks : Technologies Architectures Protocols
A. Technologies Fundamental physical technologies for telecommunications: Electronics: primarily CMOS Used for signal and data processing and storage Optics/photonics Primarily used to transport data Interface between electronics & optics is evolving as these technologies develop Advances are needed in Optical and electronic switch technologies Optical and electronic interconnects at all levels Low-power access technologies, especially wireless
B. Architectures Architectures that reduce the number of network hops Optical bypass Layer 2 rather than Layer 3 where possible Dedicated content-delivery networks
Packets Packets Packets Packets Bypass options 1. No bypass: All traffic goes to IP layer for processing All packets processed by IP router 10 nj per bit Allows aggregation of incoming traffic flows WDM Links Statistical multiplexing increases utilisation of paths IP Patch panel WDM Links Time Peak rate (channel capacity) Time IP Time Time
Bypass options (cont d) 2. Bypass TDM layer (electronic cross connect, OTN) Some traffic streams processed at TDM level 1 nj per bit TDM layer bypass IP TDM TDM WDM OXC WDM layer (optical cross connect) Some traffic switched at WDM layer < 0.1 nj per bit Switching wavelengths WDM OXC WDM layer bypass
C. Protocols Service transactions and protocols Efficiency of multi-layer protocol suite Sleep and standby states Energy-efficient Ethernet Dynamic rate adaption
Energy-efficient protocols Sleep & standby states Network devices enter low power state when not in use Can apply to systems and sub-systems Need to ensure network presence is retained Use Network Connection Proxy with sleep protocol Need to account for state transition energy and time May have multiple lower energy states IEEE Energy Efficient Ethernet (802.3az) Low power idle mode when no packets are being sent Approved Sept. 2010 Currently applies to copper interface only; not optical
Energy-efficient protocols Dynamic rate adaptation Modify capacity of network devices in response to traffic demands Change clock frequency, processor voltage Power = C Voltage 2 frequency Slower speed to reduce power consumption 100 Mbit/s uses 10-20 W less than 10 GE, 4 W less than 1 GE Need to allow transition time between rates Dynamic rate adaptation and standby states can be combined No protocol Sleep state Adaptation Both time time time time Power Packets Source: Bolla et al., 2011
Thank You