Energy Consumption in IP 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
[email protected]
Energy Consumption of the Network Hot spot
Why should we be interested in energy? • OPEX • Greenhouse Impact • Managing “Hot Spots” - Getting the energy in - Getting the heat out • Energy-limited capacity bottlenecks • Enabling energy efficiencies in other sectors
Power In
Where are We Heading ? 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
Energy Consumption Grows
Summary Modeling energy consumption of the Internet - Core, metro, and access networks Energy in network routers Energy in optical transmission Will (can) optical switching technologies help to reduce energy consumption?
What is the Carbon Footprint of Telecoms? Global Telecoms Footprint (devices & infrastructure)
Mobile Network
Mobile handsets
Fixed Narrowband
2002 20X increase
1450+% growth
2020
0 Broadband Modems
100
200
Fixed Broadband
Adapted from “SMART 2020: Enabling the low carbon economy in the information age,” GeSI, 2008 www.gesi.org
300
400
Footprint (MtCO2 p.a.)
Energy Model of Simple IP Network CRS-1 ~ 10 kW / rack
Core
Core Core
Packet over Sonet
Core
Core
Core Core
Fibre Amps
Metro
WDM 12816 Edge ~ 4 kW
Edge
OLT - 100W
Curb
Edge
Curb Curb
OLT - 100W Passive Optical Network
ONU ~ 5-10W 0.1 - 1000 Mb/s to the user Baliga et al., 2007
Curb
Access
Number of Hops in the Internet 2006 Data
0.1
Pr [ H = k ]
0.08 0.06
AT&T: 20 hops
0.04 0.02
0 0
5
15 10 Number of Hops, k
20
Source: P. Van Mieghem, “Performance Analysis of Computer Systems and Networks”, Cambridge (2006)
25
Power Consumption of IP Network 25 2007 Technology
20 router hops Contention ratio = 25
1.0
Total
15
Today’s Internet (~ 2.5 Mb/s)
Routers
10
0.5
Access (PON) 5
SDH/WDM Links 0 0
50
100
150
Peak Access Rate (Mb/s)
Baliga et al., 2007
200
250
0
% of Electricity Supply
Power (W/user)
20
Ultra-Broadband Access 100
5.0
Power (W/user)
Contention ratio = 25
Total 50
2.5
Routers Access (PON)
SDH/WDM 0
0 0
500 Peak Access Rate (Mb/s)
Baliga et al., 2007
1000
% of Electricity Supply
2007 Technology
20 router hops
Total Power Per User (W)
60
3.0
Efficiency Improvement Rate = 0% p.a
2.0
5% p.a
40
10% p.a 1.0
20
20% p.a
0
0
1
100
200
Peak Access Rate (Mb/s)
Baliga, et al, 2008, unpublished
300
400
% of Electricity Consumption
Technology Improvements
Energy Consumption in Access Networks NEC CM7710T
Access N/W Edge Node Cisco 12816
Cabinet Splitter
NEC CM7700S
Cabinet
PON NEC VF200F6
FTTN with VDSL2 Zyxel VES-1616F-34
Cisco 4503
NEC GM100
PtP Axxcelera ExcelMax CPE Axxcelera ExcelMax BTS
WiMAX
Power Consumption in Access Networks 40
Power Per User (W)
Oversubscription = 10 25 WiMAX FTTN 20 PtP PON 0 1
10
100
250
Peak Access Rate (Mb/s) • •
Wireless access consumes more energy than optical access PON FTTH is “greener” than FTTN
Network Energy Consumption per Bit 10-3 ~100 μJ/b
20 hops
Energy per bit (J)
10-4 ~1 μJ/b Total
10-5 10-6
Routers Access (PON)
10-7 WDM Links 10-8 2.5
25
250
Peak Access Rate (Mb/s)
2500
Observations •
Optical transport (WDM) consumes relatively little energy < 5% of energy > 25% of CAPEX
•
Access network dominates at low rates – Standby/Sleep mode needed
•
Network routers dominate at higher rates – Need to • reduce hop count • improve router efficiency • manage distribution and replication of content (IPTV)
Power Consumption in Routers
Power consumption (W)
1,000,000 ?
P = C2/3
100,000
where P is in Watts where C is in Mb/s
10,000 10 nJ/bit 1,000
P ~ 10 100 100 nJ/bit 10 1
1 Mb/s
1 Gb/s
1 Tb/s
Router Throughput
Source: METI, 2006, Nordman, 2007
1 Pb/s
Energy Bottleneck High-end router: Cisco CRS-1 Linecard Chassis Capacity: 0.64 Tb/s Power: 13.6 kW
Switch Fabric Chassis: Power: 8 kW
X2 every 18 months
Per Rack Source: Neilsen, 2006; Deutche Telekom, 2007
Fully equipped: Multi-rack router Capacity: 41 Tb/s Power ~ 1 MW
Electronic Routers Forwarding Engine
Fibers
J Switch Fabric Demutiplexers O/E Converters
Switch Fabrics
Buffers
Reduced bit rate (i.e. parallel processing) Electronics
Optics
Multiplexers
Energy in Electronic and Optical Routers Line Card
Data Plane
Buffer O/E
Forwarding Engine Buffer
I/O
Buffer
O/E
Forwarding Engine
Switch Fabric Switch Control
Routing Tables
Routing Engine Control Plane
Energy/bit
Power supply inefficiency Fans and blowers
Total
Electronic (2008)
0.7 nJ
3.2 nJ
0.5 nJ
1.0 nJ
1.1 nJ
3.5 nJ
10 nJ
Electronic (2018)
50 pJ
65 pJ
10 pJ
20 pJ
25 pJ
80 pJ
250 pJ
Optical (2018)
0
65 pJ
15 pJ ?
15 pJ
25 pJ
80 pJ
200 pJ
Optical Packet Switching is not a promising alternative G. Epps, Cisco, 2007, ITRS, 2005, R. Tucker, JLT, 2006
Contention Resolution in the Wavelength Domain λ1
Forwarding Engine Forwarding Engine
λn
Switch Fabric
Forwarding Engine
Fatal Flaw: Require large n for low blocking probability (n ~3 -10 x)
Power (W/user)
100
Wong, JLT 2006 Pathiban et al., JLT 2009
Total (Wavelength-domain contention resolution ) Routers 1.2 X Total (Conventional router) Routers
50
WDM 5 X
00
Access WDM 500 Peak Access Rate (Mb/s)
1000
The Challenge
Total Power Per User (W)
60
3.0
Efficiency Improvement Rate = 0% p.a
2.0
5% p.a
40
10% p.a 1.0
20 Target
20% p.a
0
0
1
100
200
Peak Access Rate (Mb/s)
Baliga, et al, 2008, unpublished
300
400
% of Electricity Consumption
10 % - 20 % p.a. continuous improvement in efficiency
Summary •
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)
-
Caching and content distribution networks
-
Continuous improvement in router efficiency
