II-EN: BlueTool: Infrastructure for Innovative Cyberphysical Data Center Management Research

II-EN: BlueTool: Infrastructure for Innovative Cyberphysical Data Center Management Research BlueTool: Infrastructure for Innovative Cyberphysical Data Center Management Research It is projected that by year 2010, the blade-based power density will reach 6 kW/ft2 [4]. In a medium-scale data center of 1,000 ft2, this means a power total of 6MW. Cooling power needs matches the computing power needs at factors ranging from 0.5 to 4, depending on the thermally efficient layout design and on the utilization of the cluster. According to a recent survey of Uptime Institute, about 46% of the data center operators do not use power-save features. The purpose of this proposal is to create a research infrastructure on cyber-physical and sustainable data center design and management that will provide i) the means to solidify the scientific evidence of savings (know-what) and the engineering to save (knowhow), and ii) a tool of awareness on the possible savings. PI Guptas IMPACT Lab and co-PI Stanziones HPCI, through NSF support (Projects 2 and 3 in Table 1) and seed grants from Science Foundation Arizona and Intel Corporation (Projects 1 and 8), have performed research on thermal-aware job scheduling and data center management. Results of this research include: i) heat recirculation can substantially increase the cooling cost (Figure 4), and ii) thermal-aware scheduling can yield more than 20% savings in total power needs (Figure 2). Co-PI Phelans Sustainable Energy Lab, through APS support (Project 6) is developing heat-activated cooling systems, which can harness the heat generated from devices such as computing servers to power the cooling process. With the evidence that considerable savings are possible, the problem is translated to reinforce the applicability and dependability of sustainable and energy-saving techniques, and increasing awareness