Reducing datacenter energy usage through efficient job allocation

A large portion of power consumed by datacenters in small and medium sized businesses is wasted consumption, as machines mostly run idle. Commonly, this is accepted by operators as many countermeasures, such as layout redesign or hardware upgrades are too. In this work we analyze how much power can be saved if an exact optimization problem is set up to determine the cost minimizing allocation of resources. Since the exact solution cannot be found in a timely manner due to the nonlinear nature of energy consumed by a system in operation, we show the goodness of various approximations in terms of power savings using numerical evaluations based on real workload traces compared to a commonly used resource allocation policy used in datacenters today.

[1]  Hendrik F. Hamann.  A Measurement-Based Method for Improving Data Center Energy Efficiency , 2008, 2008 IEEE International Conference on Sensor Networks, Ubiquitous, and Trustworthy Computing (sutc 2008).

[2]  F. Tödtling,et al.  One size fits all?: Towards a differentiated regional innovation policy approach , 2005 .

[3]  Christoforos E. Kozyrakis,et al.  JouleSort: a balanced energy-efficiency benchmark , 2007, SIGMOD '07.

[4]  Christoforos E. Kozyrakis,et al.  Automatic power management schemes for Internet servers and data centers , 2005, GLOBECOM '05. IEEE Global Telecommunications Conference, 2005..

[5]  Ricardo Bianchini,et al.  Dynamic cluster reconfiguration for power and performance , 2003 .

[6]  Patrick Kurp,et al.  Green computing , 2008, Commun. ACM.

[7]  David Vengerov,et al.  A reinforcement learning framework for utility-based scheduling in resource-constrained systems , 2009, Future Gener. Comput. Syst..

[8]  Karsten Schwan,et al.  Providing platform heterogeneity-awareness for data center power management , 2008, Cluster Computing.

[9]  Jeffrey S. Chase,et al.  Making Scheduling "Cool": Temperature-Aware Workload Placement in Data Centers , 2005, USENIX Annual Technical Conference, General Track.

[10]  Alan R. Hevner,et al.  Design Science in Information Systems Research , 2004, MIS Q..

[11]  Tajana Rosing,et al.  Temperature aware task scheduling in MPSoCs , 2007 .

[12]  Anthony A. Maciejewski,et al.  Static allocation of resources to communicating subtasks in a heterogeneous ad hoc grid environment , 2006, J. Parallel Distributed Comput..

[13]  Kirk Pruhs,et al.  Speed scaling to manage energy and temperature , 2007, JACM.

[14]  Prasad Raghavendra,et al.  Optimal algorithms and inapproximability results for every CSP? , 2008, STOC.

[15]  Karthick Rajamani,et al.  On evaluating request-distribution schemes for saving energy in server clusters , 2003, 2003 IEEE International Symposium on Performance Analysis of Systems and Software. ISPASS 2003..

[16]  R. Faure,et al.  Introduction to operations research , 1968 .

[17]  Janet L. Wiener,et al.  Cost-aware scheduling for heterogeneous enterprise machines (CASH’EM) , 2007, 2007 IEEE International Conference on Cluster Computing.

[18]  Karthick Rajamani,et al.  Energy Management for Commercial Servers , 2003, Computer.

[19]  Susanne Albers,et al.  Energy-efficient algorithms for flow time minimization , 2006, STACS.

[20]  Thomas F. Wenisch,et al.  PowerNap: eliminating server idle power , 2009, ASPLOS.

[21]  Chandrakant D. Patel,et al.  On building next generation data centers: energy flow in the information technology stack , 2008, Bangalore Compute Conf..

[22]  S. Shapiro,et al.  An Analysis of Variance Test for Normality (Complete Samples) , 1965 .

[23]  F. Frances Yao,et al.  A scheduling model for reduced CPU energy , 1995, Proceedings of IEEE 36th Annual Foundations of Computer Science.

[24]  Dror G. Feitelson,et al.  Workload Modeling for Performance Evaluation , 2002, Performance.