Zhiming Zhao, A. Belloum, M. Bubak
May 1, 2009
Citations
0
Influential Citations
19
Citations
Journal
Future Gener. Comput. Syst.
Abstract
Rapid development of Internet and Grid technologies has greatly extended the number of resources which scientists can apply for research. More and more scientific advances are achieved via experiments which consist of large volumes of computing tasks and data. Scientific Workflow Management Systems (SWMS) are a key technology to integrate computing and data analysis components, and to control the execution and logical sequences among them. By hiding the complexity in an underlying infrastructure, SWMSs allow scientists to design complex scientific experiments, access geographically distributed data files, and execute the experiments using computing resources at multiple organizations. In this way, domain scientists can effectively use available resources while focusing on the logic of the experiments instead of low level technical details. During the past few years, scientific workflow systems and applications have attracted enormous research interest. Fig. 1 shows an increasing number of workflow related publications in the journal of Future Generation Computer Systems (FGCS). A number of research foci can be found in recent publications. Running legacy applications on the Grid is essential for the domain scientists to extend execution scenarios of their experiments. Wrapping components in a scientific workflow using a standardized architecture, such as services, improves the reusability of the components and also makes the integration easy. McGough et al. [1] present a service architecture named GridSAM to manage the job submission onto Grid and tomanage different abstractions of tasks at runtime. Glatard et al. [2] demonstrate how services can be dynamically grouped and optimized in the workflow enactment and scheduling at runtime. Interoperability between different workflows is essential to realizing cooperation between different scientific experiments. Describing distributed Grid resources using semantic web technologies promotes high-level sharing of components and