Research on Novel Algorithm for Host Load Prediction

By analyzing a large of history data and predicting load variance trend of host, We can offer the important basis to load balance and decide load balance strategy for effective load balance. However most load prediction re-search focus on flow prediction of the network or load prediction algorithm it-self used classical time series prediction method, usually it is ignored on load balance after load prediction. Moreover, information exchange among nodes in distributed system through broadcast usually. It would be casting a lot of re-source of network. it is difficult to balancing load with whole system, because some node don’t know load status and change of other network nodes well and exactly. Aiming at these disadvantages, we propose a novel algorithm for up-dating load information and prediction based on mobile agent called CPLBMA . This system is based on OpenMosix parallel distributed system platform, and used series prediction method combining with mobile agent technology to im-prove the performance from a simple load prediction algorithm. According to the prediction result, using stochastic interval and hierachical cluster analysis, selecting a matched node for overload node or underload node to dispatching and migrating task, it can overcome shortage of actual prediction technology, reach load balance and the performance of CPLBMA is improved.