Research Of Job Scheduling In Data Ming Grid Base On Petri Net

In recent years, with the amount of information of data increasing sharply, data mining is faced with challenges of computation-intensive, which has become a major bottleneck for further development of data mining. As a technique of integrating resources, grid provides a method to resolve defect of computing power in data mining. Therefore, the combination of the data mining technology and grid technology has created a dedicated grid that is data mining grid. It is mainly used to solve problems of data mining for the mass of data, which requires high capacity of computing and to be able to share resources and resolve simultaneously intelligent analysis of issues of various fields in the number of grid nodes changed dynamically.Great complexity of data mining application and the heterogeneous dynamic grid cause that data mining grid has its own characteristics. So many key technologies need to be further studied and job scheduling technology is one of them. The core function of job scheduling is to identify resource requirements, to query distribute, schedule and monitor resources, to make use of resources efficiently, to improve the performance of the entire grid system and to ensure benefits of the resource users and providers. In this paper which is around the key technologies involved, we systematically and deeply make research in the structure of job scheduling, job scheduling and other aspects. The research mainly includes two aspects:Firstly, in the basis of analyzing the functional requirements of job scheduling, in this paper, we define a method for job description based on data mining grid and instructs in detail the rules of decomposition among the different kinds of granularity jobs. And in this paper, we design a job scheduling framework based on hierarchical structure and describe fully its module function, dynamic features and interactive relations.Secondly, according to structure division of job, a model of job scheduling based on hierarchical colored Petri net is designed. The model decomposes job scheduling from top to bottom. The lower the layer is, the finer granularity of job scheduling is and the upper the layer is, the stronger description of scheduling process in the overall job is focused on. It not only can define precisely the process of job scheduling, but also can analyze accurately the entire process of the job.