Load prediction using (DoG–ALMS) for resource allocation based on IFP soft computing approach in cloud computing

In today’s world, most of the applications run with the service of cloud computing, which proceeds the process using the internet. In the case of cloud computing, based on customer needs, they may increase or decrease resource utilization. Virtualization is the process of multiplexing the resources from physical machines to virtual machines. However, it is challenging to prevent overloading for each physical machine of an automatic resources management system which affects virtualization to allocate the resources dynamically. To overcome these concerns, a new algorithm is proposed in this work, which can predict the future load precisely in the physical machine and decide which may be overloaded next. Then, the necessary action is taken to prevent overload in the system. In this work, the prediction of loads for allocating future resources is presented, and the dynamic scheduling and resource allocation for the predicted tasks are performed using IFPA. The difference of Gaussian-based adaptive least mean square filter is employed for predicting the loads function points which are used to estimate the complexity and cost rate. Also, a soft computing technique (improved flower pollination algorithm) is employed for the effective resource allocation strategy. The performance of the approach is intended and compared with other conventional works. The results proved that the work has better accuracy in load prediction and provide a way to allocate the resource precisely. At the same time, the traffic at the physical machines is significantly controlled.