Carla Ellis Duke University Grand Challenge: Global Climate Change Carla Ellis Duke University It is hard to imagine that any truly "Grand Challenge" can be compartmentalized into the relatively narrow domain of distributed systems. Such problems should drive technological advances across a broad range of disciplines; just limiting the discussion to computer science and engineering, these might include new materials, circuit design, computing architecture, software systems, applications, and HCI. Certainly, distributed computing can play a role in problems significant enough to science and society to be called "grand.“ So for something sufficiently “grand” and “challenging” – global climate change / disruption Then the question is – what do D.S. have to contribute? There are two dimensions to this theme: What do Distributed Systems have to do with it? Carla Ellis / Duke University / NSF Grand Challenges in Distributed Systems Workshop @ MIT / Sept, 29-30, 2005

A “Greener” Computing Infrastructure Reducing the environmental impact of Manufacture, deployment, and disposal Energy cost of running and cooling (1) making our future computing infrastructure safer for the environment As computer scientists, we should “do no harm”. We aren’t directly spewing greenhouse gases by powering our computing infrastructure, so what’s our problem? Most of the additional electric generating capacity built in the last 5 years is oil & natural gas A recent UNU study found that the ave desktop computer with monitor requires 10X its weight in fossil fuels and chemicals to manufacture. Short lifespan – 12 million PCs landfilled annually Energy used over the lifecycle of a desktop computer > refrigerator. % of electric load <<statistics needed>> A "greener" computing infrastructure will entail (1a) reducing the energy and materials going into manufacture, deployment, and retirement of computing devices and (1b) reducing the energy (especially from non-renewable sources) of running and cooling our computing infrastructure.

Supporting Climate Science & Engineering Computational support for science of global climate change Applications to enhance energy efficiency / management outside the computing sector (2) finding ways for computing to serve the environment. Has the computer systems research community Tried to understand the needs of the climate change communities, Found an effective way to contribute our expertise, Built the right tools To have a positive impact on their endeavers? Computing to serve the environmental goals will entail (2a) computational support for the sciences studying global climate change and (2b) applications of computing to enhance energy efficiency beyond the traditional computing infrastructure.

Role for Distributed Computing Research in Greener Computing Improved lifecycle: reduce/reuse/recycle Incentive systems to encourage energy-motivated resource sharing P2P to achieve more efficient utilization of existing unused capacity Low power / low energy computing systems Broader systems context for energy management Lessons from mobile computing – tolerating disconnectedness – not always “on” Develop energy metrics / measurement expertise / tools (1a) There are estimates that the typical desktop computer requires ten times its weight in fossil fuels and chemicals in its manufacture while there exists significant unused computing capacity in the Internet. Creating better incentives (and eliminating disincentives) for sharing can allow more efficient utilization of existing resources. P2P can be viewed as an analogy to carpooling with private PCs seen as the equivalent of SUVs in every garage. Emerging technologies offering a "greener" lifecycle may require new forms of software system support. (1b) The low-hanging fruit of computing device energy management has been recently harvested. Further progress in energy efficient computing requires a more aggressive approach and a broader systems context. Carla Ellis / Duke University / NSF Grand Challenges in Distributed Systems Workshop @ MIT / Sept, 29-30, 2005

Role for Distributed Computing Research in Climate Science Deployment of sensor networks designed specifically for environmental monitoring Harvesting energy in the field Dynamic data driving large-scale scientific simulations Application-specific tools (2a) Grid computing and wireless sensor networks combine to support dynamic data-driven simulations of the environment. Carla Ellis / Duke University / NSF Grand Challenges in Distributed Systems Workshop @ MIT / Sept, 29-30, 2005

Role for Distributed Computing Research in Energy Management Managing energy distribution systems Microgrids: peer-to-peer power generation Supporting energy conservation efforts in buildings, transportation systems, manufacturing processes, etc. Interfaces exposing energy use to users Collaboration applications for more effective teleconferencing / telecommuting 2b) Distributed applications can manage energy use and distribution systems in buildings, transportation systems, homes, utilities, manufacturing processes, etc. Pervasive computing changes the way people and businesses interact with significant potential for energy savings (e.g., teleconferencing). Carla Ellis / Duke University / NSF Grand Challenges in Distributed Systems Workshop @ MIT / Sept, 29-30, 2005

Implications for NSF Inter-disciplinary research with scientists & engineers dealing with environmental and energy problems. Any new networking hardware should emphasize energy efficiency. Is it a candidate for a CISE program? Potential to redirect lots of areas of expertise in systems to new goals, different metrics & underlying assumptions. Carla Ellis / Duke University / NSF Grand Challenges in Distributed Systems Workshop @ MIT / Sept, 29-30, 2005