Analysis of Clustering Effects In Cloud Workload Forecasting

Cloud workload forecasting is becoming progressively more crucial as cloud computing gains greater omnipresence. Due to the increased user service requests the availability of finite resources in cloud computing consistently poses a significant challenge in resource management. Our aim is to analyze the need for clustering in workload forecasting in cloud computing to provide better resource management. We use various clustering methods and find the method that outperforms the others. We compare the performance metrics of various forecasting models with clustering vs the forecasting models without clustering to analyze the effects of clustering. From our results obtained we conclude that performing clustering on cloud workload dataset improves the prediction efficiency of the forecasting models