Research on Optimization Computing Service Platform Architecture

The development of network technology and parallel computation provides dramatically opportunity for complicate optimization computing research. This paper brings forward the concept of optimization computing service, which is the theoretic foundation for designing the framework of optimization computing service platform - MetaSolver. MetaSlover consist of user interface, storage service, monitor, dispatcher, executor etc. Multiple solvers and multiple tasks could be scheduled and executed cooperatively and efficiently based on dynamic load balance algorithm and its implementation is detailed. Numerical experiments conducted on MetaSolver, which are applied to different fields, demonstrates that the optimization computing service platform is feasible and effective