An Innovative Replica Consistency Model in Data Grids

In data grids, replication is a critical technology aiming at lowering the access delay as well as saving network bandwidth by duplicating original data in a distributed manner. For data-intensive applications with read/write files, replica consistency is an important issue. In this paper, we propose a novel replica consistency decision model with high adaptability and flexibility using the naive Bayesian classifier in order to improve the system performance in data grids. A system prototype has been implemented. The experimental results also prove the excellence of the proposed model.