基于双数据目录存储的网格元数据的资源发现

Looking for Resource Based on Grid Metadata Stored in Double Data Directory

提出了基于全局命名的数据网格环境下元数据目录和资源信息目录双目录存储和访问的概念,研究了全局命名的规则,给出了在该命名情况下元数据的注册和存储以及利用这种注册存储如何快速发现远程数据资源。说明了在多网格结点下寻找“对照表”采取时间戳抑制多次重发和导致多次访问数据资源的方法,并指出了“对照表”应采取的热点数据优先、冷僻数据清除的策略。

This paper brings forward the concept of storing and accessing data both in metadata directory and resource information directory based on the condition of overall situation naming, then studies the rule of overall situation naming and gives the registering and storing of the metadata. This paper also studies the problem of how to use the register and storage to find the long-distance data resource. Then, it illuminates the method of searching for the ＂comparison table＂ using the time marker in the multiple-grid-node. This method can restrain the instance of sending the same message time after time and the instance of accessing the data resource time after time. At last, the paper gives the strategy of giving priority to the common data and cleaning out the rare data.