Integrating scheduling policies into workflow engines

AbstractWorkflow applications running on distributed environments are a promising solution for resource and computing intensive problems. However, the heterogeneity of resources in these kind of environments may turn scheduling of such applications into a complicated enterprise. Although there is research in sophisticated scheduling policies for workflows they are of little actual use, as they must be ported or implemented into each and every workflow engine they intend to support. The problem becomes an m×n effort (for m policies and n workflow engines). In response to such a problem we present schedflow, a system that provides a simple interface between existing workflow engines and scheduling policies. We conducted experiments that demonstrate schedflow’s usefulness when confronted with different workflow engines and their default scheduling policies