GM-WTA: An efficient workflow task allocation method in a distributed execution environment

Abstract

A workflow is a collection of units of work called workflow tasks which cooperatively realize a business objective by utilizing system resources such as databases. The roles of workflow tasks are performed in the order driven by a computer representation of the workflow logic. During completing each task’s role, the remote control transfers and the remote resource accesses may often occur in a distributed workflow system. Hence, the efficient distribution of workflow components, especially workflow tasks, is so effective as to improve the performance of workflow processing. If we can place adjacent workflow tasks as close as possible and locate workflow tasks near to the required resources, we can significantly reduce the overhead of workflow processing. In this paper, we propose an efficient workflow task allocation method in a distributed workflow system, which is based on the locality principle. The method that utilizes the concept of graph partitioning can improve the performance of workflow processing by minimizing the remote processing costs incurred during workflow execution. In addition, we perform several experiments to evaluate our proposed method.