Performance evaluation for scientific workflow interoperability

Abstract

Scientific workflow is a special type of workflow for scientists to formalize and structure complex e-Science applications. Many workflow systems have been released to solve problems in special domains. In large collaborative projects, it is often necessary to recognize the heterogeneous workflow systems already in use by various partners and any potential collaboration between these systems requires workflow interoperability. Publish/Subscribe Scientific Workflow Interoperability Framework (PS-SWIF) approach was proposed by author to achieve workflow interoperability among workflow systems. This paper evaluates the PS-SWIF approach and its system to achieve workflow interoperability using Web Services with asynchronous notification messages represented by WS-Eventing standard. In this paper, the author presents quantitative measurements that show that PS-SWIF can scale in a workflow heterogeneous distributed environment by assessing the scalability of the system in terms of notification message loads using large number of machines and different sizes of datasets the system can manage.