CPU load prediction using ANFIS for grid computing

Effective utilization of computing resources and prediction of future resource capabilities are needed to achieve high performance computing in grid Environment. To ensure this, effective and flexible forecasting and prediction method needed to use time-shared resources for large applications which impact greater importance for scheduling. Predicting the available performance on each resource is basic problem and various prediction techniques and modeling approaches proposed in this context. The Proposed prediction approach uses the combination of Adaptive Neuro based Fuzzy Inference Systems (ANFIS) and clustering process to find the future CPU load based on the historical data. Clustering identifies the natural groupings in data from a large data set and which can be used as preprocessor for ANFIS prediction. ANFIS prediction can be applied to each and individual clusters which can show that proposed model performs better and provide minimum error results.