| Summary: | 碩士 === 立德大學 === 資訊科學與應用研究所 === 97 === Utilizing the replication techniques to create the replicas in Data Grids not only can reduce the data access costs of the clients, but also can improve the data resources reliability. Many kinds of replication strategies are proposed on Data Grids. The Simple Bottom-Up (SBU) and the Aggregate Bottom-Up (ABU) are the replication algorithms for Multi-Tier Data Grids. The ABU algorithm considers the data access relations between the clients to select the suitable data resources to be replicated, and created on the suitable sites. The ABU algorithm can reduce the data access time efficiently. However, while the data access patterns of the clients are gregarious based, the ABU algorithm cannot focus on analyzing the data access relations between the clients in this condition. In order to solve the problems of the ABU algorithm, we design a replication algorithm according to the behaviors of the tribes in the world named “Gregarious Tribe (GT) Algorithm”. The GT algorithm not only can analyze the relations of the data access between the clients as same as the ABU algorithm, but also can focus on analyzing the relations of the data access between the different tribes. According to the simulation results, the GT algorithm can reduce the data access time more efficient than the ABU algorithm. Moreover, the GT algorithm can fully exploit the storage capabilities of the sites in Data Girds.
|
|---|
