Real-time correction of performance forecast in distributed computing system resource

Effective load prediction for nodes is a good method to get resource allocation and task attempter algorithm with high performance, in order to adjust and improve performance of distributed application. Status of Grid resource varies abruptly, consequently load prediction should be corrected and adapted again, so real-time correction is introduced in the process of load prediction, which can correct the prediction algorithm from parameter estimation, error and status, so that it can implement the adaptation itself, and get better prediction precision. It shows in experiment results that using real-time correction in prediction can reduce error with little system spending.