Research On The Application Of Distributed File System In G/S Model

With the rapidly technology development, people can obtain a large number of spatial information data through various channels, and the demand for spatial information applications are increase quickly. It needs to deal with greater capacity of spatial data, faster data transfer rates and faster processing speed is also necessary. These applications put forward higher requirements for network access of spatial data. The new type of spatial data access patterns based on geo-information browser/ Distributed Spatial Data Servers (G/S), learning the advantages of C/S and B/S model, It's a new way of thinking to solve the problems faced by spatial data access. As a result of the huge amount of data access, wide variety of data formats, as well as a series of complex data organization in spatial data access, it is difficult to solve data storage and access with the database, Learn from the experience of Google File System, Distributed File System based on the G/S model can solve these problems better.This thesis originate from the project of the tutor, based on the technology of G/S distributed server clusters, to solve the application requirements of the G/S model of spatial information server clusters as the goal, on the basis of the analysis of the problems that G/S model was meet in provide spatial information services, do the following work:1. This thesis provide a viable solution for load balance and redundant storage issue with distributed servers in G/S model.2. In-depth study and comparison of NFS, Google File System, Hadoop distributed file system, FastDFS, gives a detailed comparison of the results,provide a theoretical and data support for the application of distributed file system in G/S model.3. In order to conform the application requirements of the G/S model of spatial information distribute server clusters, modifying some source code of FastDFS, Achieved use HGML management scheduling FastDFS. 4. Make use of FastDFS slove load balance and redundant storage problems successfully during release Lunar remote sensing data.