A Self-Adaptive Metadata Management Model for Data Grid

Metadata management is a key technology of heterogeneous data sources integration. Metadata management in data grid is based on distributed multi-layer mapping on resources, which has the characteristics of centralization in indexing and decentralization in resources. These characteristics may lead multiple request redirections in the relevant resource search process and eventually increase the cost of resources searching. In order to lower the probability of request redirections and improve querying efficiency, this paper studied into the metadata management model and presented an adaptive metadata management model. This model introduced the neighborhoods between metadata servers and constructed first-fit-first-link algorithm for adjusting the neighborhoods in the model dynamically. The model also replaced request redirection and its subsequent querying with the querying in metadata replication, strived to improve efficiency in resources searching.