International Journal of Cloud-Computing and Super-Computing
Volume 4, No. 1, 2017, pp 1-8
The Application of Fast Pruning Algorithm Under Map/ReduceTask Scheduling
Under the Map/Reduce, traditional task allocation and scheduling algorithm are in low overall efficiency. In order to improve the overall efficiency of the Map / Reduce for task assignment, this article use pruning algorithm which is based on the time of node processing tasks to improve the tasks assignment global efficiency. On the algorithm, system according to the specific processing time of each task in each node to quantify modeled, and establish time measurement matrix between tasks and nodes and finally using rank metric matrix pruning method to gradually reduce the size of the task assignment. For handling the N tasks assignment issues, (N-1) operations are used to obtain optimal distribution of task assignment, and it can make full use of the resources of each node. This paper uses MATLAB to simulate the algorithm and using of large amounts of data to makes a comparison between pruning algorithm and traditional scheduling algorithm. The results show that the pruning algorithm can significantly improve the overall efficiency of Map/Reduce task scheduling.