A Flexible Scheduler for Workflow Ensembles

In this paper, we propose a flexible workflow scheduler that facilitates the replacement of the objective function according to the user's needs. The possibility of replacing the objective function extends the usability of the scheduler for a variety of objectives. The proposed flexible scheduler uses Particle Swarm Optimization (PSO) to assist the production of schedules on cloud resources. We perform the evaluation via simulation using a set of scientific workflows grouped into ensembles. Three conflicting objective functions were evaluated: minimization of the overall makespan, maximization of fairness, and minimization of monetary costs. Simulation results show that the flexibility of the proposed scheduler has been achieved since each function could produce schedules that satisfied its corresponding objective.