Home – Home – Jornals – Geography and natural resources 2016 number 6S

As the information technologies are actively developing, the volume of data that needs to be processed constantly increases, which requires keeping hard- and software technologically advanced and finding new approaches to data processing. Based on the distributed computations model MapReduce, the original method of raster images processing is proposed in this paper. The MapReduce model proposes to split initial dataset into pieces using Map operation, process these data pieces and gather all results using Reduce operation. Service-oriented distributed environment of ISDCT SB RAS also faces problems of large data volumes processing, particu larly processing the raster images. In order to increase the processing speed of spatial data within the service-oriented infrastruc ture, the distribution of raster images among computational nodes was organized. Mapping of the raster images is implemented using the specifications - sets of rules of how the data should be split and gathered. The mechanism of definition and application of specifications is implemented as a part of ISDCT SB RAS Geoportal. The Geoportal allows executing distributed services in a centralized way. The use of specification during the service execution allows to effectively utilize the available computational resources. The proposed approach allows using the instruments for spatial analysis of raster images within the distributed environment without theirs modification. Execution of distributed services that work with large volumes of spatial data within the MapReduce model allows decreasing the overall services execution time and using available computing resources at higher rates.