Research

Service Architectures for Real Time Data Integration and Search

Professors Gannon and Plale

Service-oriented architectures for dynamic data driven, large-scale computational science are beginning to yield results that could directly benefit other domains that depend on efficient large-scale search and analysis for their success.

Service architectures for computational science are the research of IU Professors Gannon and Plale. The architectures are focused on the dynamic parallel and distributed execution of applications models in a framework that includes pre- and post- processing stages, responsiveness of the models to real-time environmental conditions, and resource usage ranging from large-scale computational supercomputers down to many-core server platforms.

Service support includes portal-based graphical tools for user construction of workflow task graphs, BPEL-based workflow execution engines, run-time provenance collection of both data provenance and process provenance, personal metadata catalogs that utilize the experiment nature of investigation for efficient storage and retrieval of XML. Service creation factories have been investigated for wrapping legacy codes.

Real-time continuous stream mining and event processing is provided through a service interface so event detection can be one piece of a larger workflow. This work has been funded in part by the National Science Foundation through the Linked Environments for Atmospheric Discovery (LEAD) project, the NMI OGCE project, and the Department of Energy.