云计算英文资料MRGIS A MapReduce-Enabled High Performance Workflow System for GIS.pdfVIP

MRGIS: A MapReduce-Enabled High Performance Workflow System for GIS Qichang Chen, Liqiang Wang Zongbo Shang Department of Computer Science WyGISC and Department of Geography University of Wyoming University of Wyoming qchen2, wang@ zshang1@ Abstract many programming models have been proposed. One of the most successful frameworks is MapReduce [5], which pro- The growth of data used by data-intensive computations, cesses massive datasets in parallel manner with supporting e.g. Geographical Information Systems (GIS), has far out- load balancing and fault tolerance, etc. MapReduce pro- paced the growth of the power of a single processor. The grams can be automatically parallelized and executed on a increasing demand of data-intensive applications calls for large cluster of computers. The MapReduce runtime sys- distributed computing. In this paper, we propose a high tem takes care of data partitioning, task scheduling, failure performance workflow system MRGIS, a parallel and dis- handling, and communication managing. This allows pro- tributed computing platform based on MapReduce clusters, grammers with no parallel programming experience to eas- to execute GIS applications efficiently. MRGIS consists of ily utilize cluster for data-intensive computing. a design interface, a task scheduler, and a runtime support In this paper, we propose a MapReduce-enabled high system. The design interface has two options: a GUI-based performance workflow system for applications of Geo- workflo