VECPAR'06 - Seventh International Meeting on High Performance Computing for Computational Science |
Overlapped Regions with Distributed Spatial Databases in a Grid Environment
Wladimir Meyer (Federal University of Rio de Janeiro - Brazil)
Jano Souza (Federal University of Rio de Janeiro - Brazil)
Grid computing is becoming a natural tendency to be adopted in geographic information systems (GIS). Its characteristics, based strongly in resource sharing and standardization in critical areas like interfaces, security and data transfer, attract the attention of GIS research. On the other hand the design of distributed spatial Databases (DSDB) behind this information Systems contains many kinds of challenges inherited from both distributed database and spatial database areas. The performance of queries in a weakly-coupled environment, where data is spread, is one of the most important goals to be reached. Based on the Secondo-grid [1], a framework proposed to be used as test bed for distributed spatial databases, this paper introduces some modifications to allow the exploration of parallelism when executing queries over spatial databases servers in a computational grid. A federated architecture using spatial data partitioning with some level of overlapping, and the adoption of local index structures for spatial data, are behind the environment proposed. A query broker is suggested to generate and address the subqueries to correct servers based on data locations and servers status. Web services offered by the grid middleware are used to provide reliable file transfer, database register, replica location, and automatic monitoring of resources. Every time a region of interest is stored in more than one server, the query is split into the correspondent number of subqueries to be executed simultaneously and, after the processing is done the results are transferred and combined in the requestor machine.
Data Grid infrastructures , Data Grids, Web services and P2P , Query processing and information retrieval in Grids , Replication and fault-tolerance in Grids
Logos Universidade Federal do Rio de Janeiro - Coordenação dos Programas de Pós-graduação de Engenharia Instituto Nacional de Matemática Pura e Aplicada Rio de Janeiro | Brazil | 2006 | July | 10 11 12 13