Specification of user and provenance-based adaptive control points at workflow composition level

Abstract

With the increasing capacity and power of distributed computing infrastructures in silico experiments have gained widespread popularity. The different scientific communities (physics, earthquake science, biologists, etc.) have developed their own Scientific Workflow Management Systems (SWfMS) to provide dynamic execution to the scientists. These SWfMSs differ from each other due to the divergent requirements. However, in spite of their diversity they agree on the strongest need that have not been completely fulfilled yet; runtime user-steering and adaptive dynamic execution. Additionally, when provenance data is collected during execution, provenance based steering also emerges as a big challenge. To support the scientists with special interaction mechanism during runtime we have introduced the so called iPoints, special intervention points where the scientist or the system can take over the control and are able to manipulate workflow execution based on provenance and intermediary data. In our current work we specified these iPoints in IWIR language which was targeted to provide interoperability between four existing well-known SWfMS within the framework of the SHIWA project.