The Key Techniques of Scientific Workflow System

Due to the requirements of scientific computational and data-intensive analysis, Scientific Workflow System (SWFs) technology is presented for remote resources data access and management under distributed infrastructures. The scientists are capable of using the system to address issues in their scientific domain. In this paper, firstly, we have an overview of the SWFs. Secondly, two types of scientific workflow systems are compared in the level of abstraction. And then we analyze and summarize the cross cutting issues of SWFs technology, that might be exploited in constructing workflow system effectively. We focus on the techniques, i.e., data provenance, fault tolerance and re-run. In the end, we make some suggestions towards improving SWFs.