The Analysis And Optimization Of Scheduling Algorithm In Cloud Computing Environment

With the continuous development of Internet technology, as well as expanding the range of network applications, data has become the core development of the business. Many large IT companies have set up own big data centers, while the development of enterprises are also increasingly dependent on the data, the fact that the era of big data has arrived! The Hadoop platform is emerged with the advent of the era of big data. In a cloud computing platform, resources and job scheduling strategy are the core, these plays a vital role for the whole system calculates the distribution and operation of resources. Therefore, the study of cloud computing environment scheduling algorithm is great significance.Because scheduling problems has gradually shifted from job scheduling to resource scheduling on cloud computing platform, this balance between job scheduling and resource scheduling under Hadoop platform, focusing on several existing job scheduling and resource scheduling algorithm based on Hadoop platform,including the first in first out(FIFO) scheduling, scheduling algorithm based on s load balancing and the scheduling algorithm based on resources perception and the main resource fair(DRF) to analyze the main ideas, advantages and disadvantages of these scheduling algorithms. For Hadoop platform job scheduling problem, depending on the load job make improvements scheduling algorithm based on load balancing and resource perception, propose a scheduling algorithm based on workload perception(WLWare). The algorithm is based on the cluster node information, the job is classified, by the type of work load and node load condition, the algorithm will work with the nodes match, improved resource utilization of the system. At the same time the problem of resource scheduling on Hadoop platform, the main resource for Hadoop 2.0 Fair(DRF) scheduling algorithm does not consider the indivisibility anddynamic resource usage issue a resource request is proposed based on the primary resource equity(DRF) scheduling algorithm based on dynamic resource needs more resource scheduling algorithm(DEDRF), the algorithm for the primary resource fair(DRF) algorithm introduced new resource allocation and resource status feedback factor, give full consideration to the request of the indivisibility of job resources and system resources Dynamic usage of the system nodes forward and reverse incentive to improve the utilization of resources and adaptability. In order to verify the performance of the algorithm, was validated and evaluated on Hadoop clusters and the experimental simulation platform of Cloudsim. Experimental results show that the two improved scheduling algorithm proposed can well improve the performance of existing algorithms.