数据网格中一种QoS驱动的副本选择算法

A QoS Driven Replica Selection Algorithm in Data Grid

以移动Agent作为任务载体,代表用户应用程序在分散的数据网格环境下智能地进行副本选择.利用移动Agent与各个副本宿主节点的本机通信来替代传统的远程访问方式,从而避免大量的副本传输,节省网络带宽资源.提出了一种由用户QoS(服务质量)驱动的副本选择算法,开发出基于Java Swing的实验演示系统,并针对于3种不同的QoS目标对移动Agent的迁移进行了模拟试验.实验表明提出的副本选择算法能够根据用户的不同QoS偏好来优化副本选择,当网格规模较大时更能凸显对用户QoS偏好的优化效果.

Mobile Agents act as carriers of tasks on behalf of users＇ applications,accessing replicas intelligently in the scattered data gird environment.Instead of accessing replicas remotely,mobile Agents communicate with the replica hosting nodes on the corresponding local machines,which avoids transmitting lots of replicas and saves network bandwidths.A QoS driven replica selection algorithm is presented and an experiment demonstration system is developed based on Java Swing.We have also performed simulation experiments on the migration of mobile Agents with three different QoS goals.The results show that our algorithm can optimizing the replica selection according to users＇ QoS preferences and the optimization effect will be enforced when the grid scale is larger.