Optimization of Power Usage Effectiveness in HPC Data Centers Using a Job Assignment Schedule

The use of information technologies such as high performance computing (HPC) represents a basic tool for the development of various research and production activities. The data centres in which the HPC equipment is housed represent infra-structure that supports the operation of these technologies, and the optimisation of the use of these centres has therefore become a subject of study. HPCs are based on a configuration of many computing nodes in parallel, and during their operation, different capacities are assigned to nodes for data processing; the overhead of as-signing tasks to these nodes causes a higher energy consumption by the computer HPC equipment. In this article, we examine the use of modelling for energy consumption and carry out an optimisation analysis to identify factors that generate higher energy consumption. The use of this model allows us to identify how these increase of node during processing, how they cause higher energy consumption and also increase the ambient temperature due to this use of nodes. With the aim of controlling and optimising these behaviours through modelling, we analysed how this type of processing and these temperature increases generate variations in energy consumption, and visualised this effect based on the total energy consumption of the data centre, measured based on the power usage effectiveness. We found that the total energy consumption was modified by the processing allocation schedule in these nodes using scheduling jobs assigned to HPC nodes.