Ecosystem Model Based Grid Resource Optimization Management

In this paper, the grid computing system is seemed as an ecosystem. The object of the optimization resource management is to promote the balance and evolution of the computing ecosystem. The architecture of the ecosystem model based grid resource management system is presented, which has the self-aware and selfoptimization mechanism. The knowledge discovery based self-aware mechanism has the ability to reveal the behavior patterns knowledge hide in the history grid information system. The discovered knowledge can be used to predict the resource requirement and to optimize the resource allocation. The antigen identification mechanism is studied which can identify the factors related to the computing ecosystem unbalance state. With the self-optimization mechanism of the computing ecosystem, the resource allocation problem can be abstracted as a multi-objects optimization problem. The computing expectation, ecosystem environment, and the application characteristic are considered to design policy based adaptive resource allocation and job scheduling algorithm.