A NEW FRAMEWORK OF CLUSTER-BASED PARALLEL PROCESSING SYSTEM FOR HIGH-PERFORMANCE GEO-COMPUTING

Up to now, It still remains a big challenge for us to build a high performance geo-computing system with high processing speed and also be easy of use by domain researchers. The unprecedented scale data and various complex algorithms pose many computational and management challenges To properly settle these main issues above, a new system framework for high performance gco-computing is presented in this paper A High Performance Geo-data Object Storage System (HPGOSS) base on parallel file System is used for eliminating I/O performance bottleneck and deal with the data managing problem result from the close relevancy between geo-information and remote sensing image data. Parallel programming models for fast parallelization of geo-computing algorithms are proposed. In addition, the job scheduling strategy and workflow engine are also discussed. Finally, such system could provide a parallel geo-computing environment with high performance, easy to use, optimal resource utilization, and high scalability.