Embedded In The Spatial Database Map Services, And Web Map Service Cluster Technology Research

WFS(Web Feature Service) and WMS(Web Map Service) are the two major forms of the present web geography information service. Their key difference lies in the disparate roles of the server and client in the respective architecture. As to WFS, the web browser send a request to the server which parse the request, produce search conditions, retrieve data from spatial database and send the data to clients. An application embedded in web browser must render the map and deal with the interaction between the user and the browser. But under most conditions, mass spatial data can't be transfer efficiently through network. The limited bandwidth may cause much network time delay. So, WMS is presented to solve this problem. As to WMS, the server is responsible to parse requests, retrieve spatial data which are satisfied with the conditions and render map images. Client is responsible to display the result. The map image size is limited and compressed typically, so the contradiction between mass spatial and limited network bandwidth is resolved. Due to such different mechanisms, WFS is more suitable to small data set application while WMS is a better choice for large ones.Based on a thorough study of the WMS architecture mentioned above, this article proposed the new idea of integrating the map rendering function into the spatial databse which can effectively avoid large data dumping between the map server process and the database process, thus to ultimately speed up the web map service. Meanwhile, this new method makes it possible to provide the developer with a set of user-friendly interfaces to customize the system according to their own needs.Focusing on this approach, this paper carries out the following research contents:1) A comprehensive research of the architecture of WMS in which the author elaborately analyses the role of spatial database in WMS , the relationship between spatial data management and web map service. The author points out that the spatial database functions not only as a pool to store the spatial data but also provides the basic G1S analysing functions and the project transformation function.2) This article finds that all the existing realization of WMS take the way of independent map server process which makes it unavoidable to dump large set of spatial data from database into the map server which will surely cause a cut back on the overall performance of the application, especially in case of super large dataoperation or centralized system accessing. In order to make a remarkable change, this paper present the idea of integrating the mapping function into the database by entended SQL "SELECT" command thus to get the result image directly out of the vector spatial data set.3) In order to realize vivid and diversified visual effects for the map, this paper also designs full set of map symbol data module which for the first time acchieves the symbolized espression of professional map symbol in spatial databse. During the implementation, the paper utilized the WKT format to store the symbol data which imitated the OGC Simple Feature Specification for SQL.4) Raises a new way to structure, organize and manage the spatial data in face of Internet application by self-developed functions and modules coded by process language. With the "SKIN" conception created by the author, the paper finds a simple but effective way to attach proper map legend to the proper map layer.5) Designs and develops the Linux version of GeoBeans based on spatial datacase extended with map-rendering functions6) In order to speed up the map rendering, this paper also proposes a parallel map-rendering algorithm runs on linux cluster and reserches the sever load balance during the parallel computing. In this paper, the author designed a load banlance algorithm to divide the request box to some sub-boxes. Each sub-boxes is send to a WMS node to produce some sub-maps. The result map will be reconstruct by these sub-maps. All these sub-maps are produced parallelly, so it will spend less time to pr...