Scientific Workflow Interchanging through Patterns: Reversals and Lessons Learned

Abstract

Scientific workflows are used for dealing with complex problems in different e-science domains. These workflows are modeled and executed using Scientific Workflow Management Systems (SWfMSs). Generally, SWfMSs provide their own Workflow Specification Language (WfSL), and this is a challenge considering the possibility of interchanging workflow specifications between different SWfMSs. Nevertheless, the reuse of workflows gains growing importance as it helps with fostering the collaboration and cross-fertilization across different research groups. This paper presents a research proposal, including its mishaps and assimilations, on the use of workflow patterns combined with software architecture concepts to capture the key semantics expressed in scientific workflows specified in different WfSLs and to allow the interchanging of these specifications between different SWfMSs. This paper also shows how our findings based on real world specifications led us to reformulate our initial proposal and discuss the new results.