Optimization of application execution in the GridSpace environment

This paper describes an approach to optimization of execution of applications in the GridSpace environment. In this environment operations are invoked on special objects which reside on Grid resources what requires a specific approach to optimization of execution. This approach is implemented in the GridSpace Application Optimizer
(GrAppO) which meets requirements imposed by the Grid environment as well as those specific to GridSpace model: dynamic nature of the environment, distributed sources of information, difficulty in defining suitable optimization criteria. The proposed solution applies three modes of optimization: short-, medium- and far-sighted one. The farsighted
optimization may be implemented with techniques known from the workflow scheduling research area, and for this purpose we have analyzed the possibility of using the ASKALON environment. We show that GrAppO enables to increase the quality of the GridSpace runtime performance by providing it with the most suitable objects to invoke operations on.