• Current Projects

Current Projects

• Compiler Runtime
  • We have been developing runtime and compiler support for data-intensive applications on high performance architectures. We developed a middleware system, FREERIDE back in 2001, for easing parallelization of data-intensive applications on clusters, shared memory parallel machines, and cluster of SMPs. This system has many similarities with Google's Map-Reduce, though our work predates Google's. Recently, some of the ideas from FREERIDE have been integrated with an existing implementation of Map-Reduce (Phoenix system), resulting in MATE (Map-Reduce with an Alternative API). We have also considered data-intensive applications on grids, on GPGPUs, and the emerging multi-core architectures.
• Data Management
  • Recently, we have been working on a workflow system Auscipe. Earlier, we had developed solutions for managing large-scale and low-level scientific datasets (with our approach for automatic data virtualization and automatic wrapper generation).
• Integration
  • In this area, our current emphasis is on the deep web. We have developed integrated query processing techniques on the deep web, partly motivated by the biological domain.
• Middleware
  • Our work has been driven by adaptive applications. These are applications where the user has some flexibility with respect to the final results, but there could be other constraints such as a time deadline or a resource budget. In this context, we had developed a streaming middleware system, GATES. More recently, we have considered time deadlines in the context of grid and cloud computing environments.