ZHANG Yao-nan, CHENG Guo-dong, WEI Wu-zhou, TAN Zhen-hua
A mass of characteristic data, application modes, computing programs and data analysis software has been accumulated in the long-term scientific research on resources and environments in cold and arid regions. How to effectively manage these data and integrate them with application programs, and how to provide an easy platform for scientific researchers to use these data and solve their problems with integrated models must be especially considered. In the distributed network environments, some techniques based on the Dawning 3000 high-performance computer have been studied in this paper, such as data collection and management, transplant and integration of application models and programs, data access between computing programs and database. Through testing, it is suggested that the data platform of resources and environments in cold and arid regions based on high-performance computing environments is a valuable method for improving data application and application models of cold and arid regions. The data platform consists of data center, middleware service, integrated application program service and one stop service. According to data grid concept, the general structures, the main functions and the fulfillment ways of the data platform are discussed in detail. The data in the data center, managed by Oracle database, can be divided into relation data, binary data, text data and spatial data. The middleware service, which supplies an access to the interface between data center and application service, is designed with three sections: message-oriented middleware, data access middleware, and transaction processing middleware. The application services are requested by integrated application models, application modes, and application programs such as MM5, SSIB, SME3, and ANSYS %etc%. The presentation, management, request, execution of data and their application are fulfilled thought one-stop service, which were built with three-tier architecture of Brower-Compute Service Agent-Server. The input parameters of the application software are packaged in a browser with Java Applet, and then passed to the middle-tier to form a batch of commands, which starts the application software on the server tier. On the basis of tests, a single prototype of data platform is preliminarily established based on a kind of application modes in the high-performance computing environments, and is able to serve the researchers. The data platform, managed by one-stop service, is supported by the data of the distributed environments for researching the resources and environments in cold and arid regions.