Research Of The Migration Strategy Based On T-ARG Graph In Grid

Process migration is a kind of mechanism that, according to the requirements of load balance, migrate the current active process transparently to other nodes to resume the implementation during the implementation of the task. It is an effective strategy to achieve load balance in cluster systems. As a key technology in distributed systems, the process migration has been a very active research topic at home and abroad.In this dissertation, based on the application background of engineering and scientific computation applications, we designed a migration system fitting for the related tasks. The aim is to make a reasonable allocation between the grid resource and applications,and enable these applications for optimal performance.Through studying the grid and the process migration theory, based on the results of existing research, this paper proposed a migration model which is based on related tasks of T-ARG, this model make a very good facilitating role in improving the efficiency of the implementation of related tasks. Combined with the functional requirements of related tasks, we design the overall framework of migration model of related tasks, and describe the functions of the various entities in the framework in more detail.About the migration strategy, by analyzing the migration cost of selecting the task, we proposed a migration algorithm of related tasks of T-ARG graph based on Total-Copy algorithm, and make an experiment based on this algorithm. Experimental results show that for large-scale related tasks, this algorithm has higher coordination, can better improve the efficiency of the execution of the related tasks.