Research And Implementation Of Scheduling Algorithms In Computational Grid

With developing continually, super computers have been dominant in complex science computing. Meanwhile, there are obvious disadvantage in the super-computer-centered computing model, which has to face the serious challenge. Although super computers have unimaginable computing capacity like a giant, their high design and maintain cost prevent their ubiquity, and only some national institutions, such as aeronautics and aerograph departments, could afford them. Because people need more and more complex commerce computing in our life, we have to pursue more and more powerful computing devices, but the cost of super computers is our biggest obstacle. Therefore, people start developing the low cost but high-performanc computing model. Grid computing is just one key. "A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capabilities." Computational Grid is used to connect varied computational resources to construct a virtual high performance computer, which could offer high performance computing service.Our research focuses on Task Scheduling Algorithm. Though many scholars have already done deep research to this respect, but the algorithms adopted extensively at present, such as 2- Phase, Co-RSPB, Co-RSBF, Co-RSBFR and Market-Based, etc., do not own the comprehensive abilities including guarantee of high scheduling efficiency, accurate description of resources' dynamic characteristic, evading to the negative influence brought of these dynamics, and satisfying task raisers' QoS demand. Therefore, referring to current research results, the author presents a Trust-Based and QoS-Measured Scheduling Algorithm, and then simulates it on the Computational Grid Simulation System.Our research works mainly include:Referring current research results on Computational Resource Discovery and Management Model, construct a Feedback and Investigation Based Resources Discovery and Management Modal;Referring Trust Concept in human society and research results on High-Performance Network, construct a Trust-Based and QoS-Measured Scheduling Algorithm;Referring to typical Computational Grid Realization Model, simply construct a Scheduling-Emphasized Design Model;Present 9 Performance-Evaluation measures, and then construct a Simplified Multi-Measures Performance Evaluation Model;Design and implement the Computational Grid Simulation System, and then have the varied scheduling algorithms simulated, compared and evaluated.