Massive Global Subdivision Tile Files Encoded Orderly Storage And Prefetching Technology Research

Geographic information services with data volume, number of files and more, a large number of concurrent users to access features, traditional file systems, and to Hadoop Distributed File System HDFS(Hadoop Distributed File System) as the representative of a distributed file system can not meet the massive geospatial data storage and access requirements. Therefore, geographic information systems needed to support mass storage of small files efficiently. To meet the one hundred billion small files massive scale storage and access requirements, the project team to achieve a mass-based HDFS distributed file system to support small files(SMDFS).Mapping Information System map tile data is usually a pyramid organization, data access has the characteristics of spatial locality, so if we can achieve file prefetching will effectively improve file access performance. However, the file system SMDFS thousands of small files into a single aggregated file is stored, is difficult to geographical proximity of certain files via an I/O for feedback to the user.For mapping and spatial locality of data access features and a single tile file access problem of low efficiency, this paper proposes a location-based prefetching techniques surrounding images, the purpose is to reduce SMDFS of I/O access times and improve access to the file system performance. Implementation of a prefetch prerequisite pyramid tiles file storage are ordered. This paper presents a recursive sort method quartile massive global mesh coding sequence of tiles file storage technology, making two Wiwa polymeric film to sort files based on location information, so that geographical proximity of the adjacent image storage. In the sequential storage technology, based on proposed and implemented based on an orderly pyramid prefetching techniques successfully solve massive tile reads inefficient, insufficient capacity to support concurrent access problems.