Speculation based Decision Support System for Efficient Resource Provisioning in Cloud Data Center

Cloud Computing is a utility model that offers everything as a service and supports dynamical resource provisioning and auto-scaling in datacenter. Cloud datacenter must envision resource allocation and reallocation to meet out the unpredictable user demand that touted gains in the model. This impact a need of adaptive and automated provisioning of resources aligned with clients SLA amidst the time variant environment in cloud. The robustness of dynamic resource provisioning is based on quick multiplexing virtual resources into physical servers. In this paper we put forward a prediction based automated resource allocation model induced by speculation mechanism. Our model guarantees dodging over/under utilization of resources and minimizes the cost economically without compromising Quality of Service. We regain the speculation concept that uses a past resource access pattern to predict future possible resource access. We introduce the confidence estimation factor to address the historical variability of the current pattern to improve the prediction accuracy. Experimental results show that our proposed model offer more adaptive resource provisioning as compared to heuristic and other machine learning algorithms