A scientific workflow framework integrated with object deputy model for data provenance

Abstract

There is a critical need to automatically manage large volumes of scientific data and applications in scientific workflows. Database technologies seem to be well suited to handle highly complex data managements. However, most of the workflow management systems (WFMSs) only utilize database technologies to a limited extent. In this paper, we present a DB-integrated scientific workflow framework which adopts the object deputy model to describe the execution of a series of scientific tasks. This framework allows WFMS management operations to be performed in a way analogous to traditional data management operations. Most important of all, data provenance method of this framework can provide much higher performance than other methods. Three kinds of schemas for data provenance are proposed and performance for each schema is analyzed in this paper. © Springer-Verlag Berlin Heidelberg 2006.