Prediction Model of the Running Time of Tasks Based on Load

The management of resources and scheduling computations in a grid environment is a complex and challenge undertaking,mainly due to resource’s geographic distribution,heterogeneity,dynamic,distributed ownership with different policies and priorities,varying loads,reliability,and availability conditions.Performance measurement,analysis and prediction of distributed systems have become increasingly important,especially,accurately prediction task runtime in order to achieve application performance.This paper presents a load-based task runtime prediction model which is the centre of any scheduling and resource allocation algorithm.