Research Of Multi-objective Task Scheduling Based On Fireworks Algorithm In Cloud Computing

Cloud computing is the fourth IT industrial revolution with the the development of large-scale computer and personal computer,Google first definited and developed cloud computing.As an open source model,Hadoop based on Java has ran the intensive distributed applications and analyzed the open source of distributed processing large data.The Hadoop platform has many key technologies,and the task scheduling algorithm is an important one.The operation performance and resource allocation and utilization of Hadoop platform are closely related to this technology.In this paper,through the establishment of multi-objective task scheduling algorithm based on the model of fireworks,improving the drawbacks of the three scheduling algorithms of self-systems,such as taking into account the operation time of switching costs,in the face of a large number of short operation,the processing speed of the system is not ideal.Firstly,this paper describes the system knowledge of Hadoop platform and makes a detailed introduction of the Hadoop technology,including Hadoop the two core technologies: HDFS and MapReduce,and analyzes the three kinds of scheduling algorithm with Hadoop system.Then,in this paper,in order to reduce the processing time of large-scale short operation,ensure the load balancing system,improve scheduling strategy of Hadoop's own,proposing a multi-objective task scheduling scheme solved by fireworks algorithm,to avoid the frequent switching of short operation time.At the same time,applying the MapReduce parallel programming model is used to improve the existing serial fireworks algorithm,and accomplishing the parallelization of the multitask scheduling based on fireworks algorithm,which greatly improves the operation efficiency of the platform.It can be seen from the experimental results,the serial experiment of fireworks algorithm is smaller in the range of optimal fitness value.With the expansion of the population scale of fireworks,the execution time of the algorithm does not increase suddenly,significantly better than the PSO algorithm and GA algorithm.The result of the parallel algorithm of fireworks experiment shows that,as expansion of the population scale,operation time is not increased,and it stables at a certain value.The cluster advantage is very obvious,which is more suitable for mass data processing.In the test of the speedup ratio of the cluster,the execution time of the job decreases with the increase of the cluster nodes,which can significantly improve the system's processing capability.