Supporting Dynamic Workflow Adaptation in a Dataflow-Constrained Workflow Net

Dynamic workflow adaptation problems have been recognized by the workflow community for a long time and several approaches have been proposed so far. Most of the existing approaches are mainly to guarantee a workflow instance can be propagated to a new workflow specification and can terminate properly without causing structural errors, however, for data-centric scientific workflows, correctness of data flows is also crucial to execution of scientific workflows. Although some researches have identified several data flow issues, however, data flow changes and their influence on running workflow instances have been neglected. In this paper, we first present a dataflow-constrained workflow net, and give a simple yet effective algorithm of model checking to guarantee both the structural and the data flow correctness of the modified model. Then we propose a correct transfer criterion to determine the compliance of a workflow instance with a changed workflow specification and state the conditions of transferring existing instances to the changed workflow specification. Finally, for those instances compliant with a new workflow specification, we give a transfer rule to guarantee these instances can be changed to a correct state and terminate properly.