Greenscale Home Page

As computing and data storage increasingly become a globally-available, public utility, the proliferation of large numbers of servers and massive data centers will have a substantial energy footprint in our future. The typical server consumes as much energy in one year as an SUV. Worldwide, businesses now spend $30 billion to power their data centers, and that cost is growing rapidly. Energy expenditures are already becoming more significant than the cost of machines, making energy efficiency of critical importance to our future information technology infrastructure and natural environment.
The goal of the Greenscale Center is to leverage key strengths at UCSB to face the new millenium’s energy challenges. Energy-efficient computing can not be achieved without the interplay between computer science, electrical engineering, mechanical engineering, and environmental science. Designing tomorrow’s large-scale computing systems will require a vertically-integrated effort to drive key energy-efficient technologies in computing, electronics, and building systems. Collectively, these technologies address very significant near-term and long-term energy challenges and their impact will require evaluation in economic and environmental terms.
 
 

Emerging Technologies for Energy-Proportional Computation

Today’s servers are optimized for high energy efficiency only when used at peak or when at idle. Unfortunately, servers are often used at 30-50% capacity, in particular in large datacenters where medium loading is necessary in order to provide stability in the presence of workload bursts or machine failures. What the industry needs is energy-proportional computing, where energy efficiency remains high regardless of the load on a server. Microprocessors already have the technology to provide this energy proportionality through frequency and voltage scaling. What is needed is redesign of all other aspects of server system, from network hardware, to tertiary storage, to memory, to internal busses. Optical and capacitive coupled signaling technology promise to significant gains for both the inter- and intra-machine communication. Solid-state non-volatile storage offers an attractive alternative to mechanical disks.
UCSB has extensive research in CAD and circuit techniques to reduce energy consumption (Cheng, Marek-Sadowska), with funding from MARCO-DARPA and NSF.  Finally, UCSB also has extensive research in architectures for emerging technologies (Chong, Sherwood), funded by NSF. Research in emerging technologies is heavily influenced by the available cooling technologies, as many devices have different characteristics at different temperature regimes. Furthermore, evaluating the combination of cooling and computing technologies requires an overall life-cycle analysis of the energy or carbon footprint of the manufacture, use, and disposal of both parts of the combination.
 
 

Cooling Technologies

As computational facilities scale, the flow and mixing of forced air cooling becomes a complex optimization problem. Greenscale works closely with the Institute for the Dynamics of Energy Efficiency and its Applications, which focuses on many of these problems. Research at the Institute into airflow and control algorithms (Mezic, Hespanha, Khammash, Moehlis, Matthys) is extensively funded by AFOSR and DARPA. Further research into aggressive cooling techniques includes titanium MEMS / microfluidics for thermal ground planes (Meinhart), which produces heat sinks 500 times more efficient than copper. Even more aggressive is the use of nanofluidic channels to convert heat back into energy (Pennathur). Critical to much of this research in cooling technologies is the ability to effectively model computationally the dynamics of these systems. As a result, existing collaborations (Mezic and Gilbert) already focus on efficient numerical algorithms for modeling airflows on supercomputers and we expect new collaborations to focus on specialized architectures (Gibou, Chong, Sherwood) and the use of sensor data (Suri) to calibrate reduced-order models (Moehlis).
 
 

Energy-Aware Computation

Key to energy-efficient computing is the restructuring of computations to be more energy aware. Virtualization technology (Wolski, Krintz) is a powerful tool with which to migrate and consolidate computations when used in conjunction with models and control of cooling technologies. Database techniques must be re-optimized not only for performance, but also for energy efficiency (Agrawal,El-Abbadi). Once again, a critical component of this research is the coupling of computation to the other disciplines – cooling, emerging technologies, and life-cycle analysis.
 
 

Life-Cycle Analysis and Policy

Since our energy-efficiency solutions involve the interaction of many technologies, both electronic and mechanical, we need a framework from which to evaluate their overall energy or carbon footprint, as well as economic and policy mechanisms to encourage their use. Such a framework can be found in life cycle assessment (LCA) according to ISO 14040/44 (2006), the most mature and widely used method for quantitative and comprehensive environmental assessments of product systems. LCA quantifies the environmental impacts of product system across all its life cycle stages and across a wide range of impact categories. It is thus ideally equipped to determine the environmental benefits of energy-efficient computing strategies. This approach is a generalization of a previous collaboration between computer science and environmental science (Chong, Geyer), which focused on microchip reuse. Another critical issue is the economics of the energy-efficient strategies. Market-based, self-sustaining energy-saving strategies need to be profitable for the involved economic agents. Economic assessments of these strategies are thus necessary to gauge their market potential. To be profitable business propositions, strategies for adopting new technologies and practices may require novel business models, revenue structures, business relationships and supply chain contracts. There is now a significant body of literature dealing with the business aspects of closed-loop supply chains which will inform and guide our research and to which, in turn, we will contribute. Our aim is to identify and evaluate the business models required to profitably implement encourage adoption of energy-saving technologies. Such business models for energy efficiency need to be formulated and studied in the context of all pertinent laws and regulations, since they can significantly impact the feasibility and profitability of energy strategies. The EPA’s upcoming Energy Star rating system for servers and networking hardware are particularly relevant.
Greenscale works closely with the UCSB Bren School for Environmental Science and Management, which has several faculty (Geyer, Dozier, and Frew) active in LCA, policy, and economic research. Active research in these areas has been funded by the UK Engineering and Physical Research Council, the Royal Society of Wildlife Trusts, the International Iron and Steel Institute and the California Resources Agency.
 
 

Wireless Networking

As wireless networks witness widespread deployment worldwide, the collective energy consumption of these devices is on target to skyrocket. Greenscale researchers work to curb this energy consumption through access point scheduling (Almeroth, Belding), energy-efficient protocols (Almeroth, Belding, Rodoplu), and spectrum scavenging (Zheng).
Although networks with a high-density of wireless access points provide users with capacity, flexibility, fault-tolerance, and mobility, one recent study has shown that many such WLANs are rarely utilized at their peak capacity. Our preliminary work has demonstrated that we can cluster APs based on coverage area, and power down large percentages of APs when the network is not heavily utilized. We seek to continue this work, to investigate new, more flexible means of energy savings when networks are under-utilized.
In parallel with this work is our collaboration with the Meraka Institute in South Africa. In this partnership, we seek to develop mesh networking solutions suitable for the constraints of rural, under-developed regions of Africa. A first-class requirement of these solutions must be that our solutions are power efficient, and are resilient to power outages. Our work will seek to develop protocols and higher layer solutions that are energy efficient, while still providing basic connectivity to the citizens within these regions.
 
 
Administration 
 
Name
 Title
Fred Chong
 Director
 
Affiliated Faculty 
Name
 Group(s)*
Agrawal, Divy
 EAC
Almeroth, Kevin
 WN
Banerjee, Kaustav 
 ETEPC
Belding, Elizabeth
 WN
Bianchini, Ricardo
 Rutgers University
Brewer, Forrest
 ETEPC
Cheng, Tim
 ETEPC
Dozier, Jeff
 LCAP
El Abbadi, Amr
 EAC
Frew, James
 LCAP
Geyer, Roland
 LCAP
Gibou, Frederic
 CT
Gilbert, John
 CT
Gurumurthi, Sudhanva
 University of Virginia
Hespanha, Joao
 CT
Khammash, Mustafa
 CT
Krintz, Chandra
 EAC
Marek-Sadowska, Malgorzata
 ETEPC
Matthys, Eric
 CT
Meinhart, Carl
 CT
Mezic, Igor
 CT
Moehlis, Jeffrey
 CT
Pennathur, Sumita
 CT
Rodoplu, Volkan
WN
Sherwood, Timothy
ETEPC, CT
Suri, Subhash
CT
Wenisch, Thomas
University of Michigan
Wolski, Richard
EAC
Yang, Tao
ETEPC
Zheng, Haitao
WN
 
*Groups Key
 
Abbreviation
Corresponding Group
ETEPC
Emerging Technologies for Energy-Proportional Computation
CT
Cooling Technologies
EAC
Energy-Aware Computation
LCAP
Life-Cycle Analysis and Policy
WN
Wireless Networking