Cloud Computing Environment Of Gis Spatial Analysis Task Scheduling Strategy Research

With the continuous improvement of technologies and means to obtain spatial informations, much more spatial data we could get. How to make full and effective use of these spatial data is the universal concerned question of academic field and commercial world.Parallel computing is an effective method to improve the processing speed of massive tasks.The performance of parallel computing system largely depends on its task scheduling. A good task scheduling will not only reduce the response time of the task and increase system’s I/O, but also improve the resource utilization of the entire cluster. Therefore, the task scheduling research of GIS spatial analysis in parallel computing environment has very important practical significance.Cloud computing is a popular parallel computing. The open source distributed parallel programming framework of cloud computing named Hadoop which has implemented MapReduce parallel programming model has been widely used in the present.In this thesis, on the basis of existing research, study was carried out on the subject of GIS spatial analysis task scheduling in the Hadoop platform. The main research work of this paper was illustrated as follows:About parallelization processing of GIS spatial analysis:By taking the application of spatial analysis which based on the spatial analysis operator named distance for example, this paper realized parallel GIS spatial analysis by using MapReduce parallel programming model.About Task scheduling of parallel GIS spatial analysis:This paper summaried the characteristics of task scheduling in cloud computing environment, and analysed the strengths and weaknesses of Hadoop task scheduling policies which are native or common. Then, put foward an improved task scheduling policy of FIFO which based on data locality.At the end of the paper, a Hadoop platform for experiments was set up. GIS geometry objects were expressed with WKT format, and GIS datatype was represented as a single line in a Comma-Separated-Values (CSV) file in order to be applicable to the Hadoop platform. By the experiments, the performances of dealing with GIS spatial analysis on Hadoop and the improved task scheduling algorithm were validated. Experimental results showed that Hadoop fares better for large GIS spatial analysis jobs than single-computer environment does, and the improved FIFO task scheduling algorithm the paper proposed can improve the data locality of the node, thus reduce the communication cost, shorten the average response time and the overall runtime.