Study On Implementation Of Geocomputation Job Process And Job Scheduling

GeoComputation is an important advanced computing paradigm in GeographicInformation Science. It is an indispensable part of the next generation of GIS(Geographic Information System). GeoComputation involves lots of processes such asdata processing, spatial analysis, process simulation, etc. The process ofGeoComputation is so complicated that usually consumes large amount of time. HPC(High Performance Computing) provides a powerful tool to solve the complexGeoComputation problems. In HPC environment, in order to meet user’s demand forspecific complex problems, it often combines resources and geographical jobs whichhave different functions with business process logic. It forms GeoComputation jobprocess. In order to learn how the GeoComputation job process operate in HPCenvironment correctly and efficiently, this dissertation study on the following aspects：To cope with the parallel implementation issues of the GeoComputation jobprocess in HPC environment, this dissertation proposed a modeling method of theGeoComputation job process which describe jobs and the dependencies between jobsclearly. we proposed an parallel execution framework for GeoComputation job process.The framework relying on HPC environment provides a solution for GeoComputationjob process operating correctly and efficiently.Secondly, this dissertation designed a method for processing concurrent processesbased on multithread. As the core technology of the executing engine for theGeoComputation job process, this technique provides support for parsing processmodel correctly, parallel execution geographical job and full use of system resources.The experimental results proved that the proposed method can effectively modelGeoComputation job process with good practical performance.Thirdly, this dissertation designed the time prediction algorithm with historicalinformation based on the characteristics of GeoComputation jobs and apply thisalgorithm in the backfill scheduling algorithm to improve the schedulingperformance.Then the paper designed the job scheduling strategy for GeoComputationjob process. The experimental results proved that the strategy reduce the overallresponse time of the process effectively and improve the utilization of system resources.Finally, a prototype system of parallel executing the GeoComputation job process inHPC environment is designed and implemented, which can automatically resolve anddispatch the GeoComputation job process. The experiment validated that all keytechnologies and solutions discussed in this dissertation is correct and effective.Thesystem has good practical performance, which can be widely used in geospatial data application fields.