Issue No. 05 - September/October (2005 vol. 7)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MCSE.2005.91
Beth Plale , Indiana University
Dennis Gannon , Indiana University
Yi Huang , Indiana University
Gopi Kandaswamy , Indiana University
Sangmi Lee Pallickara , Indiana University
Aleksander Slominski , Indiana University
Large scientific collaborations that use Grid technology often do so because they must conduct complex data analysis and computational experiments requiring wide-spread resources in distributed and remote locations. These experiments may involve complex workflows that run for days. The LEAD project (Linked Environments for Atmospheric Discovery) is a collaboration between meteorologists, computer scientists, and educational experts to construct a large-scale service-oriented architecture that is capable of responding to weather phenomena in real time, executing multi-model simulations of weather forecasts on demand across distributed Grid resources, and adapting resource allocation dynamically in response to the results. At the heart of the system is a triad of services cooperating to ease the increasingly onerous burden on the scientist of managing the data products used in and generated during the process of computational experimentation. The first, a workflow system, is capable of dynamic control of experiment execution, the second, a metadata catalog, actively manages an individual's experiment history over time and engages with the workflow engine to organize products so that later searching can be done with more ease than current solutions allow. The third component is a notification system that serves as the underlying communication substrate. This paper describes the three services in detail with emphasis on the interactions between the services that are needed to accomplish the active capture and recording experimental products.
grid computing, data mining, metadata, Web services
A. Slominski, S. L. Pallickara, Y. Huang, G. Kandaswamy, D. Gannon and B. Plale, "Cooperating Services for Data-Driven Computational Experimentation," in Computing in Science & Engineering, vol. 7, no. , pp. 34-43, 2005.