Research On Resource Organization And Discovery Mechanism In Meteorological Data Grid

With the rapid development of information technology, meteorological informatization of our country witnessed three times fast develops period during the90’s of last century. Nowadays, the amount of meteorological data produced by lots of observation devices such as remote sensing and automation is increasing by geometric series. Therefore, meteorological departments are trying to search the solution to manage those huge meteorological data, in order to provide data sharing for the scientific researchers who are separated geographically. The appearance of data grid has solved the problem of resource management and sharing. So it will be appropriate to apply the data gird technology in the field of meteorology. The researches on grid resource organization and resource discovery will be valuable due to its important role in grid resource management.For the lack of flexibility and scalability in resource organization and the weakness of inefficiency and load imbalance in resource discovery, our paper makes a research on resource organization and resource discovery. The concrete research contents and results are as follows:(1) Reviews the existing researches on meteorological grid. Then we classify and discuss the advantages and disadvantages of existing grid resource organization models. According to the analysis of meteorological nodes distribution in grid system and the resource description of meteorological data based on metadata, we abstract a meteorological resource organization based on hybrid architecture. In addition, we make a comprehensive introduction on the2-layer structure which consists of the super-node layer structure and logic resource tree structure in a region, including member protocol and the process of nodes register.(2)Analyses and discusses the existing researches on grid resource discovery mechanism. Then we abstract a meteorological resource discovery mechanism based on the hybrid resource organization model. We propose an improved learning strategy through using the pre-request mechanism. This new strategy will be applied to the resource discovery on the super-node layer. Meanwhile, we introduce the resource discovery process in a region using the resource mapping information which maintains in attribute manage nodes. At last, we explain the whole resource discovery process in data grid and evaluate the resource discovery algorithm.(3)Makes a comprehensive introduction on the simulate experiment of our resource organization model and resource discovery algorithm based on SimJava toolkit, including the introduction of SimJava and specific classes in the experiment. Then, we focus on the analysis of experiment results with the experiment graphs.