A Prediction-based Two-Stage Replica Replacement Algorithm
Tian Tian, Junzhou Luo · 2007
To access large and widely distributed data on data grid quickly and efficiently is an important goal of the implementation of data grid. Due to high latency of the Internet, large amounts of data need to be replicated in multiple copies at several distributed sites. However, the storage capacity is limited. So a good replacement algorithm is important to the efficiency of the access to the replicas. In this paper, we propose a prediction-based two-stage replica replacement algorithm. This algorithm achieves a good balance between value and cost by predicting replica value to make sure which replica will be replaced, and predicting the replacement cost to make it as low as possible. Simulation results show that compared with traditional replacement algorithms our prediction-based two-stage replica replacement algorithm shows better performance and efficiency of the data access on data grids.