Analysis Of Service Scheduling And Resource Scheduling Based On Cloud Computing

The generation of cloud computing has accelerated the development of the Internet industry, users obtain on-demand customized services, the cloud computing promote the transformation of service-oriented and the service has been provided greater flexibility. The service scheduling and resource allocation in cloud environment have a major impact on the whole performance of computing, it could map the service and resource that user needed,reduce job execution time, increase system efficiently, and capable of making rational resource scheduling to improves cluster utilization.Hadoop is an open source distributed computing for large data analysis platform in cloud computing. Map/Reduce in Hadoop is a programming model of cloud computing,massive Map/Reduce cluster commonly used in the treatment of PB-level data, and therefore the effect of the implementation of the data is very important. Based on this, we study the Map/Reduce theory, proposed a stratified scheduling model under Hadoop, test and verify the feasibility and effectiveness of the algorithms.First, it discussed the cloud service scheduling and resource scheduling theory. We carried out a detailed study and analysis of Map/Reduce's definitions and methods in Hadoop.We compared the advantages and disadvantages of existing cloud computing scheduling mode. It provided support for the theory behind the scheduling model.Second, it approached a stratified scheduling method based on Hadoop. This method is characterized by intensive and computationally intensive, divides the tasks to determine job priority, and through data locality and overall task completion rate allocate resources. This method could solve the over or under-utilized resource issues and provide services for more users.Finally, it verified the feasibility and effectiveness of the stratified model. It rewrites the configuration files of Hadoop platform to accommodate the proposed algorithm. By using service files in Hadoop platform as data set, it proves the feasibility of stratified scheduling algorithms, and test further demonstrated by comparing the effectiveness of the proposed scheduling algorithm.