zInternet Systems and Storage Lab

Trends towards consolidation and higher-density comput-ing congurations make the problem of heat management one of the critical challenges in emerging data centers. Conventional approaches to addressing this problem have focused at the facilities level to develop new cooling tech-nologies or optimize the delivery of cooling. In contrast to these approaches, our paper explores an alternate dimen-sion to address this problem, namely a systems-level so-lution to control the heat generation through temperature-aware workload placement. We rst examine a theoretic thermodynamic formulation that uses information about steady state hot spots and cold spots in the data center and develop real-world scheduling algorithms. Based on the insights from these results, we develop an alternate approach. Our new approach lever-ages the non-intuitive observation that the source of cool-ing inefciencies can often be in locations spatially un-correlated with its manifested consequences; this enables additional energy savings. Overall, our results demon-strate up to a factor of two reduction in annual data cen-ter cooling costs over location-agnostic workload distri-bution, purely through software optimizations without the need for any costly capital investment.