Evaluation of multiobjective swarm algorithms for grid scheduling

Often, solutions to complex problems are found in nature. Swarm algorithms are capable of solving such complex problems by implementing patterns from nature. This patterns are found in a variety of scientific fields. In this paper, we discuss two swarm algorithms extracted from Biology and Physics, namely: Multiobjective Artificial Bee Colony (MOABC) and Multiobjective Gravitational Search Algorithm (MOGSA). The first one is based on bees behavior and the other follows the gravity between masses. These algorithms are implemented to solve the grid scheduling problem. Optimization of job scheduling is one of the most challenging tasks in Grid environments because it severely affects the execution time of an experiment (set of jobs). Experiments often are tied up to fulfill deadlines and budgets. One of the main contributions of this work is adding multiobjective processes to these swarm algorithms to minimize those conflictive objectives. Results show that MOABC clearly improves the MOGSA approach when solving the problem. MOABC is also compared with real grid meta-schedulers as Deadline Budget Constraint (DBC) and Workload Management System (WMS) by using the simulator GridSim to prove the improvement that offers this new algorithm.