| Grid computing, regarding as a new technology for next generation computing is characterized by its resource sharing when be used for solving large-scale problems in many fields. In recent years, researches on grid computing are mainly focuses on scheduling of grid resources and it is becoming a prospection subject. because resources in grid are dynamics, diversity, and autonomy.In grid computing systems, grid resources distributed on geography are owned by different organizations with different access polices, cost considerations and loads capabilities. The service providers (resource owners) and consumers (resource users) have different goals,objections,and requirements so that some traditional resource management and scheduling algorithm are not suitable here to bring economical models to grid computing system and find a way to utilize these models on them,have becoming a new research field in grid computing.In this thesis, resource scheduling algorithm based on DAG model and econometrics model were analyzed extensively. For DAG model, GS algorithm and its improved algorithm called OGS were discussed based on these algorithms, cost-time optimization scheduling algorithm is proposed. This algorithm reduces the costs compared with OGS algorithm without increasing the completion time at the same time. On the other hand, based on deadline budget constrain algorithm for econometric model and considered the characters of grid resources and the demands of the users for grid computing resources. A cost-time compound scheduling algorithm based on cost functions is provided. The algorithm not only achieves better performance in budget than the algorithm based on deadline when they finished the same number of gridlets with the same time,but also gets a lower compltion time than the algorithm based on budget when they finished the same set of gridlets with the same budget.The experiment results show that these new optimization algorithms presented in this thesis are feasible. |
PDF Full Text Request