Energy Saving Policies in Grid Computing and Smart Environments

1 Energy-Saving Policies in Grid-Computing and Smart Environments Alejandro Fern´ andez-Montes Department of Lenguajes y Sistemas Inform´ aticos Universidad de Sevilla, Sevilla E-mail: [email protected] This work studies the problem of energy consumption growth in two spheres: Grid-Computing and Smart Environments. These problems are tackled through the establishment of energy-saving policies developed for each environment in order to save the maximum energy as possible. In the Grid-Computing environment, seven energypolicies were designed in an attempt to minimize energy consumption through shutting resources down and booting them. It is proved that approximately 40% of energy can be saved. Efficiency of various grid locations was compared using Data Envelopment Analysis (DEA) methodology. In Smart Environments where sensors perceive lighting conditions, the energy-saving policy adjusts lighting in order to satisfy user preferences and prevents energy from being wasted. A set of wireless sensors were deployed on two offices at the department of Computer Languages and Systems. The dataset created over several months was employed to extract information about user lighting preferences, from the application of which it is proven that around 70% of energy can be saved in lighting appliances. Keywords:, efficiency, grid-computing, DEA, energy policies Introduction Energy efficiency and sustainability are two areas of great interest in recent decades. CO2emissions worldwide continue growing significantly. From 1990 to 2009 CO2emissions rose by approximately 38% while population grew about 28%. With respect to Information Technologies, IT energy consumption represents a momentous 3% to 5% of CO2emissions worldwide which is similar to that of aviation transport. The huge amount of energy consumed by grid-computing provides justification for a study into energy-saving methodologies. To this end, grid operational policies must be analyzed in order to be optimized. An experimental Grid organization, located in France and called Grid’5000, was analyzed in this work. The first relevant contribution is creating energy policies that establish the managing of grid resources. While other research attempts to reduce the makespan, the policies shown in this work aim to describe and compute what to do with a resource when it finishes the execution of a job. Thus, each energy policy decides to either leave resources switched on or switch them off depending on the purpose of the policy. The second relevant contribution is to compare energyconsumption efficiency between each Grid’5000 location, where productive efficiency is measured as the energy consumed to run Grid’5000 jobs using DEA methodology which was introduced in [5]. With respect to Smart Environments, the research provided in this work, adapts the utilization of new technologies in order to improve consumption by adapting lighting conditions used in daily activities to user lighting preferences. Contributions Energy policies for Grid-computing Energy policies establish the managing of grid resources by describing what to do with a resource when it finishes the execution of a job. Thus, each energy policy decides to either leave resources switched on or switch them off depending on the purpose of the policy. Seven energy policies were developed and the results of applying them are shown in [2]. Two of them are based on probabilistic distributions (Exponential and Gamma) while five of them are deterministic and based on the state of the grid. Results in Table 1 show that Always off is the best policy in terms of energy saving. However, it is the policy which forces the highest number of power cycles, and hence the stress on the hardware is the greatest. Statistical-based policies, Exponential and Gamma, perform reasonably well in terms of the percentage of prediction success, and hence, energy saved by each shut down is the greatest, and the AI Communications ISSN 0921-7126, IOS Press. All rights reserved 2 Energy Boot. Energy Simulated Saved Policy + consumed vs per Name Shutt. MWh curr shutt. Alwz Off 1,668K 149 68.59% 0.04 Load 1,569K 150 69.01% 0.04 S.Off Ts 653K 150 69.11% 0.10 Exp. 1,047K 151 69.67% 0.06 S.Off Rdm 890K 157 72.38% 0.07 Gamma 183K 186 85.82% 0.17 Alwz On 0 217 100.0% 0.00 Table 1 Selected results, sorted according to Energy Consumed. stress imposed on the hardware is the lowest. Efficiency comparison of locations with DEA In this work [3], DEA is used as a method to compare energy-consumption efficiency between each Grid’5000 location, where productive efficiency is measured as the energy consumed to run Grid’5000 jobs at each location (comparison of Energy Policies was also performed). Bordeaux 100% Lille 100% Lyon 100% Nancy 65% Orsay 63% Rennes 68% Sophia 57% Toulouse 95% 0% 20% 40% 60% 80% 100% 120% Scale efficiency VR Scale Efficiency Comparison by Locations Fig. 1. Comparison of locations VR Scale Technical Efficiency In the analysis of locations, it can be observed that Bordeaux, Lille and Lyon are the most efficient locations (Variable Return to Scale Tecnical Efficiency (VRSTE) equals to 1.000 for these policies), followed by Toulouse, and that the least efficient locations are Sophia and Orsay, followed by Nancy and Rennes. In terms of dispersion, Bordeaux, Lille and Lyon have the most homogeneous behaviour between policies, followed by Sophia, with Toulouse being the location whose performance is the most dispersed between policies, followed by Sophia, Rennes and Nancy. Figure 1 shows this graphical comparison of VRSTE per locations. Adjust of Lighting in Smart Environments In this work [4], it is shown that electricity consumption can be reduced by incorporating two devices (a regulator of light and a motion sensor) in an indoor setup. Hence, any saving obtained in electricity consumption can be greater if the preferences of users and the environment are set up by following an optimal criterion. Thus, we proved that in a experimental room, the waste of lighting is about 72% of consumption in spite of the optimality of the geographical orientation. Furthermore, the methodology proposed is adaptable to any environment and any user, due to the fact that lighting conditions are not predefined but learned by a smart environment system. Days computed 365 Days with lights on 68.77 Total Luxes generated 598,775,280 Total Luxes wasted 429,210,527 KWh CO2(kg) Euros Total Consumption 475.32 130.24 64.24 Total Waste 340.72 93.36 46.05 Table 2 Light consumption and wasting per year for single office. Making a smart adjustment of the lights could save about 340 kWh, 93 kg of CO2, and 46 (0.14/kWh) a year per standard office, based on the data retrieved for the case study and the methodology proposed. See Table 2 for an analysis of the results. Acknowledgements This research is supported by the Spanish Ministry of Economy and Competitiveness ARTEMISA (TIN2009-14378-C02-01) and Simon (TIC-8052) of the Andalusian Regional Ministry of Economy, Innovation and Science. References [1] Fern´ andez-Montes, A. Energy-Saving Policies in Grid-Computing and Smart Environments. PhD thesis, Universidad de Sevilla, Spain, 2013, available at: www.lsi.us.es/pop/repositorio.php [2] Fern´ andez-Montes, A., Gonzalez-Abril, L., Ortega, J.A., Lefvre, L. Smart scheduling for saving energy in grid computing. Expert Systems with Applications, 39(10), 2012. [3] Fern´ andez-Montes, A., Velasco, F., Ortega, J.A. Evaluating decision-making performance in a grid-computing environment using DEA. Expert Systems with Applications, 39(15), 2012. [4] Fern´ andez-Montes, A., Gonzalez-Abril, L., Ortega, J.A., Velasco, F. A study on saving energy in artificial lighting by making smart use of wireless sensor networks and actuators. IEEE Network, 23(6), 1620, 2009. [5] Charnes, W., Cooper, W. W., Rhodes, E. Measuring the efficiency of decision making units. European Journal of Operational Research, 2, 429444, 1978.