Polyrhythmic Harmony Search for Workflow Scheduling

In this paper we propose the new conception of Harmony Search (HS) algorithm for scientific workflow scheduling called Polyrhythmic Harmony Search (PHS). The main idea of Harmony search is based on the artificial phenomenon found in jazz musical performance, namely the process of searching for better harmony. As it well known, jazz is characterized by the presence of several complex musical structures, such as polyrhythm, where a few different rhythmic patterns are played in the same time. According to technology of virtualization, which allows to divide physical computing resources into virtual machines, the main idea of the proposed algorithm is in the use of two heterogeneous harmonies – task scheduling on the computing environment and optimization of the computing environment for the tasks – as two different rhythms, which can be played in jazz. This algorithm is compared to several metaheuristic algorithms. Experimental results of PHS evaluation showed that proposed algorithm allows to find an optimal solution on a par with Genetic algorithm and with a greater convergence speed in terms of scheduling execution time.