ABSTRACT
Workfl ow Scheduling ........................................... 240 9.4 Scientifi c Workfl ow Runtime Scheduling
and XML Stream Optimization ................................................. 244 9.4.1 What is Optimization? ....................................................... 245
9.4.1.1 Static Optimization and Runtime Optimization .......................................................... 246
9.4.1.2 Semantic Optimization ......................................... 246 9.4.2 Execution of Optimized Scientifi c Workfl ow ................ 247
9.5 Conclusions and Future Work .................................................... 248 References ............................................................................................ 248
e-Science is a buzz word when it comes to connecting different kinds of sciences and communities with each other to share scientifi c interests, data, and research results. This connection is the trend of scientifi c and technological development that augurs a rapid increase in the number of computations being employed by e-scientists. Consequently, scientifi c workfl ow, a new special type of workfl ow often underlying many large-scale complex e-science applications such as climate modeling, structural biology and chemistry, medical surgery, or disaster recovery simulation, deserves intensive investigation. Compared with business workfl ows, scientifi c workfl ow has special features such as computation, data or transaction intensity, less human interaction, and a large number of activities. Some emerging computing infrastructures such as grid computing, with powerful computing and resource sharing capabilities, present the potential for accommodating those special features. Some work both theoretically and empirically has been done toward this research frontier such as GredbusWorkfl ow, Kepler, Taverna, and SwinDeW-G series [1]. Each piece of work highlights different aspects of scientifi c workfl ow with different emphasis.
