Design Of Task Scheduling Algorithm In The Distributed Computing Platform

With the rapid development of science and technology in the fields of academic research and engineering practices, through a large number of complex calculations required to achieve the goal. These are on the computer's performance made high demands, but high-performance supercomputers practical choice is unrealistic. Therefore, proposed a scheme for complex tasks or decomposition of large quantities of data appropriate to assign these decomposition fragments to smaller computing resources distributed over a plurality of computer calculations, finally calculate the final result, so that the calculation of the free resources are used effectively.Firstly, introduces the basic architecture of distributed computing platforms. Analysis and comparison of common distributed computing model to study the three-tier logical architecture of distributed computing systems, proposed the establishment of a client-based- on the server side- three-tier structure of distributed computing node model of computing platform design framework; computing platform provides a simple interface to achieve the application of this interface can be used to build a computing network system to accomplish complex computing tasks on client computing platform, server, compute nodes detailed design.Then make one for the distributed computing platform based on adaptive Min-Min scheduling algorithm to improve resource classification, taking into account the shorter task completion time and computing resource load balancing, its detailed description and by using simulation software GridSim experiments carried out to verify the feasibility analysis algorithm.Finally, the implementation of distributed computing platform, the platform for verification work done, indicating that the feasibility and correctness of the computing platform.